In November, ICE launched a process to get private sector “skip tracing” services, where corporate investigators use digital snooping tools and on-the-ground surveillance to track immigrants in exchange for monetary bonuses. ICE procurement records indicate the agency will be targeting as many as 1.5 million immigrants in the U.S.

Related

Lawmaker Challenges ICE Plan to Hire Bounty Hunters

Taken together, records show the 10 companies have made over $1 million to date — and stand to make over $1 billion by the contract’s end in 2027. Some of the companies’ roles in the bounty hunting program have been previously reported, including by The Intercept, but others — such as Bluehawk, EnProVera, and Gravitas — are being revealed here for the first time. (None of the 10 companies commented for this story.)

The bonanza for federal contractors comes as the Trump administration’s focus on deportations has led to a massive increase in ICE’s budget. The companies range from those with extensive experience doing intelligence work to those with more mundane government contracting experience, like finding janitors for federal agencies.

Among the companies poised to cash in on the bounty hunting program, the largest potential haul — over $365 million — could go to Capgemini Government Solutions, a McLean, Virginia-based federal consultancy that has a long track record working for the departments of Defense and Homeland Security, including providing intelligence services for ICE.

Florida-based Bluehawk LLC stands to reap the second largest payout from bounty hunting, at over $200 million. Bluehawk is a longtime contractor for the Pentagon and U.S. intelligence community, providing intelligence collection and analysis, as well as counterintelligence services.

In September, Bluehawk announced it was beginning counterintelligence work for the Department of Homeland Security. Like some of the other contractors tapped by ICE, the company is focusing on immigration after honing its capabilities doing war on terror-era military and intelligence operations. The company’s advisers include former Defense Intelligence Agency chief Ronald Burgess and Dell Dailey, a retired Army lieutenant general who ran U.S. Joint Special Operations Command following the September 11 attacks.

Government Support Services, a contractor that helps staff roles for janitors, groundskeepers, and security guards at agencies across the federal government, could make upward of $55 million on the bounty hunting program.

EnProVera, another company signed up for a contract on the bounty hunting program, also boasts a broad range of federal contracts. The company advertises a variety of intelligence-gathering and investigative services on its website, promoting its work with Customs and Border Protection.

EnProVera CEO Larry Grant’s past work experience includes “conducting clandestine overseas operations, authoring a highly classified study of a foreign nation’s technical capabilities,” and “architecting intelligence systems support to combat operations,” according to his biography page.

EnProVera could make nearly $3 million by the contract’s end.

Constellation Inc., which has previously landed administrative contracts across the Department of Homeland Security, is looking at a potential $58 million payday from bounty hunting.

SOS International, or SOSi, another experienced military contractor, landed a bounty hunting contract around when the program was revealed in November. SOSI’s skip-tracing work for the program, which was first reported by The Lever, could earn the firm up to $123 million. The company is a longtime military contractor whose past work spans operating a major military base in Iraq to operating overseas propaganda campaigns. SOSi’s website notes the company uses large language models in its government work.

Other firms have more traditional private investigative backgrounds.

Gravitas Investigations, which could make over $32 million through the bounty hunting contract, says it offers “comprehensive surveillance operations.” The company touts its skill at locating anyone using a combination of digital sleuthing and real-world tracking.

Related

Deportation, Inc.

“We go where your Subject goes,” its website says. “We follow on foot, in a vehicle, onto public property, and anywhere legal. Our surveillance operatives covertly document your Subject’s activities with a handheld, high-definition camcorders, and covert cameras.”

Gravitas says it makes extensive use of social media and other online data to pinpoint individuals on its customers’ behalf.

The company Fraud Inc. “strives to validate our clients’ suspicions,” according to its website, using a variety of public and private databases, social media digging, and video surveillance. “We also can obtain legally high-altitude video,” it boasts.

Among the more novel firms on the bounty hunting contract is AI Solutions 87, whose role was recently reported by 404 Media. The company is providing “AI agents” to ICE that it says can autonomously track “people of interest and map out their family and other associates more quickly.”

Perhaps the most provocative bounty hunting firm is BI Incorporated, an immigrant-tracking subsidiary of GEO Group, the for-profit prison giant whose fortunes have rapidly climbed following Trump’s reelection and the funding boom for deportation operations. With lucrative contracts for both hunting and imprisoning immigrants — its bounty hunting work could net $121 million by 2027 — GEO Group now stands to generate revenue through multiple stages of the administration’s ongoing deportation campaign.

The post 10 Companies Have Already Made $1 Million as ICE Bounty Hunters. We Found Them. appeared first on The Intercept.

ICE has secured a deal with surveillance firm BI Incorporated as part of a new program, first reported in October by The Intercept, to use private bounty hunters to determine the locations of immigrants in exchange for monetary bonuses.

BI, which was acquired by the GEO Group in 2011, is one of several firms hired by ICE to provide “skip tracing” services, in which its teams of corporate investigators will use surveillance to track immigrants across the country to their homes and places of work so federal agents can easily swoop in and make arrests.

Records show ICE has already paid BI $1.6 million, with the potential for the contract to grow to as much as $121 million by the time it concludes in 2027.

ICE’s push to privatize its hunt for immigrants has drawn the scrutiny of Rep. Raja Krishnamoorthi, D-Ill., who warned it “invites the very abuses, secrecy, and corruption our founders sought to prevent.”

Neither BI Incorporated nor GEO Group immediately responded to a request for comment.

The deal illustrates a strategy of vertical integration within GEO Group, which has found a growing line of business operating for-profit immigration detention centers under the second Trump administration. In this case, the corporation stands to be paid by the federal government to both find immigrants and then to imprison them.

Related

Deportation, Inc.

Shares of GEO Group, which donated to both Trump’s reelection campaign and his inaugural fund, spiked following his 2024 victory. Trump’s return to office has proven fortuitous for GEO Group: The president’s “Big Beautiful Bill” earmarked $45 billion for jailing immigrants. “This is a unique moment in our company’s history,” GEO Group CEO J. David Donahue told investors in May, “and we believe we are well-positioned to meet this unprecedented opportunity.”

GEO Group has faced decades of criticism over alleged mismanagement of its facilities and claims of rampant abuse of inmates. In August, The Intercept reported the suicide of a Chinese immigrant held at a GEO Group-operated prison in Pennsylvania. The American Civil Liberties Union filed a federal complaint over the facility in July, criticizing “horrific conditions” at the prison, including repeated instances of medical neglect.

In 2023, GEO Group was hit by a class-action lawsuit alleging the “months-long poisoning” from a chemical disinfectant of more than 1,300 inmates at a California immigration detention center. In May, Tufts University student Rümeysa Öztürk, jailed for her criticism of the Israeli genocide in Gaza, alleged her GEO Group-managed jail delayed treatment while she experienced an asthma attack.

Related

ICE Plans Cash Rewards for Private Bounty Hunters to Locate and Track Immigrants

The ICE contract record does not say whether BI would provide on-the-ground bounty hunting services, software-based investigative services, or a combination of both. ICE has previously told potential bounty hunting contractors, “It is dependent upon the vendor to complete the work required by contract,” but “Should a vendor choose to subcontract, that is at their discretion,” according to procurement correspondence reviewed by The Intercept.

BI has a long history in immigrant surveillance, having received hundreds of millions of dollars from the government to date through past contracts for ankle monitor-based tracking. The company specializes in remote surveillance and person-monitoring services, including sales of GPS bracelets and other tracking devices. “Location tracking enables individuals to work and live in the community while being monitored closely for curfews, movement, and more,” according to the company’s website. “BI offers ankle bracelet, wrist-worn, and mobile tracking solutions to meet the needs of varying risk levels.”

BI also touts its suite of software products, including case management applications for monitoring the movements of immigrants and other targets, as well as tools that allow agencies to chart a target’s “geographic and spatial location data” across Google Maps. It is unknown if the company has access to commercial mobile device locational data, or relies solely on body-mounted trackers.

But with many years of detailed GPS data pertaining to the every movement of hundreds of thousands of immigrants, BI and GEO Group hold a trove of locational information that would be of obvious value to the bounty hunting initiative.

In a November contracting document pertaining to the skip tracing effort, ICE told potential bounty hunting vendors they are “expected to provide their own internal skip tracing tools,” providing contractors with a great deal of latitude to employ surveillance products and techniques of their choosing. The document further noted that private ICE bounty hunters will not be provided credentials to identify them as agents of the government.

404 Media reported Thursday that ICE had also contracted with AI Solutions 87, “a company that makes ‘AI agents’ to rapidly track down targets.”

The post ICE Hires Immigrant Bounty Hunters From Private Prison Company GEO Group appeared first on The Intercept.

Mentions of democracy are noticeably absent, however, from Anduril’s recent announcement of a new joint venture with a state-run bomb maker from an authoritarian monarchy that is facilitating a genocide.

Anduril is partnering with EDGE Group, a weapons conglomerate controlled by the United Arab Emirates, a nation run entirely by the royal families of its seven emirates that permits virtually none of the activities typically associated with democratic societies. In the UAE, free expression and association are outlawed, and dissident speech is routinely and brutally punished without due process. A 2024 assessment of political rights and civil liberties by Freedom House, a U.S. State Department-backed think tank, gave the UAE a score of 18 out of 100.

The EDGE–Anduril Production Alliance, as it will be known, will focus on autonomous weapons systems, including the production of Anduril’s “Omen” drone. The UAE has agreed to purchase the first 50 Omen drones built through the partnership, according to a press release, “the first in a series of autonomous systems envisioned under the joint venture.” The Omen drone was described as a “personal project” of Anduril founder and CEO Palmer Luckey, a longtime Trump ally and fundraiser.

EDGE Chair Faisal Al Bannai explained in a 2019 interview that EDGE was working to develop weapons systems tailored to defeating low-tech “militia-style” militant groups.

The UAE has been eager to sell its weapons around the world, both to generate profit and to exert political influence. This most recently and brutally includes Sudan, where the Emirates supply the Rapid Support Forces, an anti-government militia. Weapons furnished by the UAE have been instrumental in the ongoing civil war, now widely described as having descended into an RSF-perpetrated genocide. In October, video imagery emerged from Sudan showing RSF soldiers indiscriminately slaughtering civilians in Darfur. Reports of rape, torture, and other atrocities at the hands of the RSF are now widespread, and a current “low estimate” of people murdered by the RSF during its recent takeover of the Sudanese city of El Fasher is 60,000, according to a recent report by The Guardian. The Trump administration determined in January that the RSF’s massacres constituted a genocide, echoing assessments by the Biden administration and human rights observers.

The RSF has been able to rapidly overtake the Sudanese army with the help of weapons from Anduril’s new partner. An April investigation by France 24 found EDGE subsidiary International Golden Group funneled tens of thousands of mortar rounds into Sudan for use by the RSF.

Nathaniel Raymonds, who leads the Humanitarian Research Lab at the Yale School of Public Health, told The Intercept mortars were among “three weapons systems that went into the hands of RSF that changed the course of the war.”

Raymonds, whose office at Yale previously partnered with the State Department to monitor atrocities in the Sudanese civil war, described Anduril’s joint venture as “mind-boggling” given the role Emirati drones and other weapons have played in facilitating the RSF’s genocide. “You have a DIA and [State Department] assessment that in a just world will trigger Leahy Act and shut this thing down from day one,” Raymonds said, referring to legislation that nominally prohibits the provision of assistance to foreign militaries that have committed major human rights violations.

Related

Trump’s Big Beautiful Gift to Anduril

Neither Anduril nor EDGE Group responded to a request for comment. A November press release from both companies noted “EDGE and Anduril will work closely with U.S. and UAE authorities to ensure full compliance with applicable laws and regulations including trade compliance rules and regulations.”

A 2024 report by Human Rights Watch noted the use of drone-delivered thermobaric bombs sold by EDGE. In October, The Guardian reported the RSF’s use of armored personnel carriers manufactured by an EDGE subsidiary. In 2024, a United Nations panel of experts deemed the UAE’s backing of the RSF as “credible,” and this year a bipartisan group of U.S. lawmakers issued a statement criticizing “[f]oreign backers of the RSF and SAF–including the United Arab Emirates.” The Wall Street Journal reported in October that both the State Department’s intelligence office and the Defense Intelligence Agency agreed the UAE was supplying the RSF with a wide array of weapons, vehicles, and ammunition. The UAE has repeatedly denied this support despite ample evidence.

Jean-Baptiste Gallopin, a researcher at Human Rights Watch who has tracked the flow of arms into Sudan, told The Intercept that EDGE Group’s products have exacerbated the horror of the ongoing war. “The rapid support forces, which we found responsible for crimes against humanity across Sudan, has made widespread use of armored vehicles made by Nimr, a subsidiary of Edge Group,” he said. “The name of Adasi, another subsidiary of Edge Group which specializes in drone technology, appeared on crates of Serbian-made 120mm munitions that the RSF has been using and which equip some of their quadcopter attack drones.” Nan Tian, a senior researcher at the Stockholm International Peace Research Institute, added that the Nimr vehicles are armed with “a gun that is made by KNDS which is a French-German arms maker. KNDS has a military partnership with EDGE Group.”

Raymonds argued that “not since Operation Cyclone,” the CIA effort to arm the Afghan mujahideen, “has there been a covert action by any nation state to arm a paramilitary proxy group at this scale and sophistication and try to write it off as just a series of happy coincidences.”

EDGE was launched at a 2019 inauguration ceremony overseen by Mohamed bin Zayed Al Nahyan, the crown prince of Abu Dhabi, and consists of over 30 subsidiaries spanning bombs, drones, ammunition, and various military and intelligence software systems. EDGE’s chair of the board, Faisal Al Bannai, is a businessman and adviser to the prince.

EDGE isn’t the only Emirati weapons company, but the conglomerate represents the bulk of the country’s arms industry by volume and illustrates the amorality of its export policy, according to Sam Perlo-Freeman, a researcher with the Campaign Against the Arms Trade, which has advocated for an arms embargo against the UAE. “As a state-owned company, they will be used as an agent of Emirati state policy,” he said. “Arms supplies to allies and proxies across the Middle East, North, and East Africa has been for quite a while a major facet of Emirati state policy.” This has manifested beyond furnishing arms to the RSF, with the UAE arming militaries in Libya, Somalia, and the ongoing genocidal war in Tigray. “There’s very few conflicts in the in the wider region that the UAE haven’t had a hand in, and very often a rather malign hand.”

Reports of EDGE wares winding up in the hands of armed proxies stretches back over a decade.

A 2013 report by the United Nations Security Council found International Golden Group facilitated the import of hundreds of thousands of rounds of ammunition into Libya in violation of a global arms embargo.

In 2019, a report by Arab Reporters for Investigative Journalism found UAE-backed combatants in the ongoing Yemeni civil war armed with pistols manufactured by Caracal, an EDGE subsidiary.

As in Sudan, a nominal civil war waged within the Tigray region of Ethiopia was exacerbated by foreign entanglement and a flood of outside weaponry. In 2023, Gerjon’s Aircraft Finds, an aviation analysis Substack, published imagery indicating the import of guided bombs manufactured by Al Tariq, another EDGE subsidiary, for use by the Ethiopian Air Force, responsible for widespread civilian death during the Tigray war.

Anduril, most recently valued by private investors at over $30 billion, has a wide array of weapons in the U.S. and with its allies, including Australia and Taiwan. It works closely with the Department of Defense and has operated surveillance towers along the U.S.–Mexico border for nearly a decade. Its business has surged as it has cast its products as a vital tool in a tech arms race between the West and China, matching the company’s rhetoric positioning it as a lethal bulwark against autocracy.

Related

OpenAI’s Pitch to Trump: Rank the World on U.S. Tech Interests

Luckey has long cast his company as a defender of democracy. “Soldiers who defend western values should all be superheroes with superpowers,” he tweeted in 2019. In an interview that year, Luckey explained backing democratic allies against “rogue nations” around the world: “I like working with the British,” he said. “Everyone’s a little bit different but more or less we all believe in western values and democracy and universal human rights.”

Anduril co-founder Matt Grimm similarly advanced the company’s moral case for an arms race on human rights grounds, describing China in a 2024 interview as the world’s “greatest evil,” denouncing the Chinese state’s “basic approach to human rights.” Grimm added that “I think they’re conducting an ongoing genocide with their Uyghur population, I think their approach to free speech, to political speech, to religious freedom, are fundamentally antithetical to how the West values human life and how we think about human rights.”

“The fact of Anduril saying they’re an arsenal of democracy and partnering with EDGE Group, it’s obviously ridiculous,” said Perlo-Freeman, “but it’s part of the broader picture of Western democracies treating the UAE as a valued partner and ally and shielding them from consequences.”

The post Anduril Partners With UAE Bomb Maker Accused of Arming Sudan’s Genocide appeared first on The Intercept.

As gossip, official leaks, and RUMINT (a portmanteau of rumor and intelligence) about a coming war with Venezuela reign in Washington, Defense Department contracting documents reviewed by The Intercept offer one of the most concrete indications of the Pentagon’s plans for operations in the Caribbean Sea over the next three years.

The contracting documents earmark food supplies for almost every branch of the U.S. military, including the Coast Guard, Army, Navy, Air Force, and Marine Corps. They detail an effort by the Defense Logistics Agency, or DLA, to source “Fresh Bread & Bakery products to Department of Defense (‘DoD’, or ‘Troop’) customers in the Puerto Rico Zone.” One spreadsheet outlining supplies for “Puerto Rico Troops” notes tens of thousands of pounds of baked goods are scheduled for delivery from November 15 of this year to November 11, 2028.

Foodstuff set to feed the troops include individually wrapped honey buns, vanilla cupcakes, sweet rolls, hamburger rolls, and flour tortillas.

The Pentagon has built up a force of 15,000 troops in the Caribbean since the summer — the largest naval flotilla in the Caribbean since the Cold War. That contingent now includes 5,000 sailors aboard the USS Gerald R. Ford, the Navy’s newest and most powerful aircraft carrier, which has more than 75 attack, surveillance, and support aircraft.

The surge of combat power comes as the U.S. has conducted more than 20 strikes on suspected drug smuggling boats in the Caribbean Sea and Pacific Ocean, killing more than 80 civilians. As part of that effort, the Trump administration has secretly declared that it is engaged in a “non-international armed conflict” with 24 cartels, gangs, and armed groups including Cártel de los Soles, which the U.S. claims is “headed by Nicolas Maduro and other high-ranking Venezuelan individuals,” despite little evidence that such a group exists. Experts and insiders see this as part of a plan for regime change in Venezuela that stretches back to Trump’s first term. Maduro, the president of Venezuela, denies that he heads a cartel.

Mark Cancian, an analyst with the Center for Strategic and International Studies, told The Intercept that the documents suggest the outsized American military presence in the Caribbean could continue for years.

“The procurement’s length of time and the level of effort seemed to point to these operations continuing at the current level for several years,” said Cancian, who previously worked on defense procurement at the Office of Management and Budget. “That’s significant because it means that the Navy will maintain a large presence in the Caribbean that is far larger than what it has been in recent years. It further implies that the Navy will be involved in these counter-drug operations.”

The Pentagon has tried to keep the details of its military buildup in the region under wraps, failing to answer questions from The Intercept about troop levels, the bulking up of bases, and warships being surged into the Caribbean. “For operational security reasons, we do not release itemized operational details of asset, unit, and troop movements and locations,” said a spokesperson for Southern Command, which oversees military operations in the region. “Information released is published via official communication web sites and social media accounts, or shared with reporters via news releases and updates.”

Related

Secret Boat Strike Memo Justifies Killings By Claiming the Target Is Drugs, Not People

The Trump administration has deployed at least 13 warships, five support vessels and a nuclear submarine — including the Ford, which is the largest vessel of its kind — to the region since August. This ramp-up includes three guided-missile destroyers: the USS Jason Dunham, the USS Gravely, and the USS Stockdale. Adm. Alvin Holsey, the outgoing SOUTHCOM commander, recently visited the amphibious assault ship USS Iwo Jima, which has been operating in the Caribbean for months. The Iwo Jima Amphibious Ready Group includes the Iwo Jima; amphibious transport dock ships; and the 22nd Marine Expeditionary Unit, or MEU, a unit especially skilled in amphibious landings.

One DLA document lists as recipients of the food an array of U.S. naval vessels known to be involved in ongoing buildup of troops and vessels including the Iwo Jima, Fort Lauderdale, San Antonio, Jason Dunham, Gravely, and Stockdale, as well as the special operations mothership MV Ocean Trader, which makes periodic appearances at hot spots around the world. The list also mentions the USS Truxtun, a guided missile destroyer not previously reported as part of the Caribbean naval buildup.

As the troops have flooded into the region, the quantities of food and costs listed in the contracting documents have mushroomed.

The initial contracting documents, released in August, included cost estimates and an estimated deliverable quantity of food linked to three locations in Puerto Rico. These were revised in September and October. Hanna Homestead of the National Priorities Project, who analyzed the documents for The Intercept, noted that the final amendment, released on October 9, included a cost estimate that increased 40 percent from the original request. The amount of food, measured in pounds, also skyrocketed 450 percent, she observed. And the number of locations in Puerto Rico jumped from three to 16.

“Those specific ships will be rotated in the months ahead, but they are likely a placeholder for the level of effort,” Cancian added. “As these ships leave, the assumption is that others will replace them. One of the questions we hope the new National Defense Strategy answers is whether this larger Caribbean deployment is long term. This food order seems to imply that it is, though the regional logistical command may just be preparing for a higher level of demand, without being sure whether the new strategy will dictate that.”

Another former defense official, who spoke on the condition of anonymity due to his current job with a military contractor, said that the documents raise significant questions that the Defense Department would rather not address. “People will ask whether this means escalation from the strikes on smugglers into a Venezuelan campaign, whatever that eventually looks like,” said the former official who has significant experience in military logistics, procurement, and supply chains.

Other locations in Puerto Rico named in the DLA documents include Muñiz Air National Guard Base within Luis Muñoz Marín International Airport; Fort Buchanan, a U.S. Army installation near San Juan; and Roosevelt Roads naval base. The latter, a Cold War-era facility previously dormant since 2004, is listed as hosting Marines. The base, roughly 500 miles from Venezuela, began receiving Marine Corps aircraft and roughly 4,500 Marines in early November.

A September 4 amendment noted “the Delivery Schedule will include one (1) additional customer. They are as follows: DoDAAC – M20179, Customer – USS Hiroshima.” The Hiroshima is a fictional warship that exists only in the “Star Trek” universe. But Homestead, of the National Priorities Project, pointed out that the Defense Activity Address Code M20179 corresponds with the 22nd MEU, according to a Fiscal Year 2026 Marine Corps logistics document.

Troops from the 22nd MEU are currently conducting training exercises in Trinidad and Tobago, the Caribbean island nation only miles from Venezuela. Maduro called the drills “irresponsible” and said the neighboring country was “allowing their waters and land to be used to gravely threaten the peace of the Caribbean.” Members of the unit have also conducted reconnaissance and surveillance training at Camp Santiago in Puerto Rico.

For months, the 22nd MEU has failed to respond to The Intercept’s questions about its operations in the region. The unit also did not respond to recent repeated requests for comment about its use of Defense Activity Address Code M20179 and the potential for food deliveries into late 2028 for troops in and around Puerto Rico.

The DLA documents are also no anomaly. Other recent contracting documents detail “food catering services for 22d MEU personnel located at José Aponte de la Torre Airport, Puerto Rico, from 15 September to 31 December 2025.” The Defense Logistics Agency is also looking into a separate “potential six-month contract for full-service food support to visiting U.S. Navy Ships” in Puerto Rico. That deal would include foods from beef steak, chicken cutlets, and lasagna to chocolate pudding, brownie mix, and chocolate chip cookie dough, not to mention breakfast burritos with bacon, egg, and cheese.

Last week, Defense Secretary Pete Hegseth said that the campaign of attacks in the Caribbean and the Pacific is called Operation Southern Spear. Led by Joint Task Force Southern Spear and Southern Command, “this mission defends our Homeland, removes narco-terrorists from our Hemisphere, and secures our Homeland from the drugs that are killing our people,” he wrote on X. Southern Spear kicked off earlier this year as part of the Navy’s next-generation effort to use small robot interceptor boats and vertical take-off and landing drones to conduct counternarcotics operations.

Trump recently teased the possibility of holding talks with Maduro; Maduro said he is open to face-to-face talks with Trump.

The Pentagon has reportedly presented Trump with various options for attacking Venezuela, according to two government officials who spoke to The Intercept on the condition of anonymity because they were not authorized to disclose information from classified briefings. Pentagon press secretary Kingsley Wilson did not reply to a request for comment.

Related

War in Venezuela, Brought to You By the Same People Who Lied Us Into Iraq

Trump has also publicly spoken of moving the sea attacks to land, confirmed that he secretly authorized the CIA to conduct covert operations in Venezuela, threatened future attacks on Venezuelan territory, and said he has not ruled out an invasion of Venezuela by U.S. troops. Asked if the U.S. was going to war against Venezuela, Trump nonetheless replied: “I doubt it. I don’t think so.” But when asked if Maduro’s days as president were numbered, Trump replied: “I would say yeah. I think so.”

White House spokesperson Taylor Rogers did not reply to questions from The Intercept about plans to attack Venezuela, the options for strikes presented to Trump, and the contracting documents which indicate the U.S. will have a major troop presence in the Caribbean into late 2028.

“These documents suggest that the Trump administration plans to maintain a significantly increased military presence in the Caribbean through the remainder of President Trump’s term in office. With ongoing military strikes against alleged drug traffickers in the Caribbean and the Pacific, the potential for escalation between the U.S. and Venezuela in particular is high, even if the administration isn’t seeking it,” Gabe Murphy, a policy analyst at Taxpayers for Common Sense, a nonpartisan budget watchdog advocating for an end to wasteful spending, told The Intercept.

The post U.S. Military Documents Indicate Plans to Keep Troops in Caribbean Through 2028 appeared first on The Intercept.

This case hinges on a recent effort by the Trump administration to unmask Instagram and Facebook accounts monitoring immigration agents in Montgomery County, Pennsylvania. It’s not the first effort of its kind by federal authorities.

In 2017, The Intercept reported an attempt by U.S. Customs and Border Protection to reveal the identity of the operator of a Twitter account critical of President Donald Trump by invoking, without explanation, its legal authority to investigate the collection of tariffs and import duties. Following public outcry and scrutiny from Wyden, the Department of Homeland Security rescinded its legal summons and launched an internal investigation. A subsequent report by the DHS Office of Inspector General found that while CBP had initially claimed it needed the account’s identity to “investigate possible criminal violations by CBP officials, including murder, theft, and corruption,” it had issued its legal demand to Twitter based only on its legal authority for the “ascertainment, collection, and recovery of customs duties.”

The report concluded that CBP’s purpose in issuing the summons to Twitter was unrelated to the importation of merchandise or the assessment and collection of customs duties,” and thus “may have exceeded the scope of its authority.” The OIG proposed a handful of reforms, to which CBP agreed, including a new policy that all summonses be reviewed for “legal sufficiency” and receive a sign-off from CBP’s Office of Professional Responsibility.

Eight years and another Trump term later, CBP is at it again. In October, 404 Media reported that DHS was once again invoking its authority to investigate merchandise imports in a bid to force Meta to disclose the identity of MontCo Community Watch, a Facebook and Instagram account that tracks the actions of immigration authorities north of Philadelphia. A federal judge temporarily blocked Meta from disclosing user data in response to the summons.

In a letter sent Friday to DHS Secretary Kristi Noem, Wyden asked the government to cease what he describes as “manifestly improper use of this customs investigatory authority,” writing that “DHS appears to be abusing this authority to repress First Amendment protected speech.”

The letter refers to the 2017 OIG report, noting that CBP “has a history of improperly using this summons authority to obtain records unrelated to import of merchandise or customs duties. … The Meta Summonses appear to be unrelated to the enforcement of customs laws. On the contrary, DHS apparently is trying to expose an individual’s identity in order to chill criticism of the Trump Administration’s immigration policies.” Wyden concludes with a request to Noem to “rescind these unlawful summonses and to ensure that DHS complies with statutory limitations on the use of 19 U.S.C. § 1509 going forward.”

Related

The Feds Want to Unmask Instagram Accounts That Identified Immigration Agents

The MontCo Community Watch effort followed an earlier attempt this year to unmask another Instagram account that shared First Amendment-protected imagery of ICE agents in public. This subpoena, first reported by The Intercept, focused not on merchandise imports. Instead it invoked law “relating to the privilege of any person to enter, reenter, reside in, or pass through the United States,” even though the subpoena was issued pertaining to “officer safety,” not immigration enforcement.

DHS did not immediately respond to a request for comment

The post Wyden Blasts Kristi Noem for Abusing Subpoena Power to Unmask ICE Watcher appeared first on The Intercept.

Unfortunately for Patel, tracking flights is legal, easy, and an important tool of government transparency.

The location of aircraft within and around the United States is public because the law requires it to be: The Federal Aviation Administration mandates that aircraft must be trackable for safety reasons, namely, to prevent them from crashing into each other all the time. Aircraft, whether privately owned or operating for a major carrier, from a small prop plane to a jumbo jet, are generally required by law to carry a radio transmitter, called a transponder, that continuously broadcasts its GPS coordinates and other information, such as altitude and ground speed. Thanks to what’s known as the Automatic Dependent Surveillance–Broadcast, or ADS-B, unencrypted transponder signals are received by other planes in the sky and anyone with a compatible antenna on the ground. According to the FAA, “ADS-B improves safety and efficiency in the air and on runways, reduces costs, and lessens harmful effects on the environment.”

Accessing these broadcast coordinates from the ground is only slightly more complex than tuning in to a local radio station. Entire online communities of aircraft hobbyists, researchers, journalists, and others make use of this open source data to chart the travel of foreign dignitaries, military movements, corporate executive trips, and, now, the director of the FBI.

As vessels of the rich and powerful, the ability to track private flights provides undeniable value to the public: It’s been used to monitor Russian oligarchs, map the CIA’s foreign torture program, and calculate Taylor Swift’s carbon footprint. That this tracking is entirely legal hasn’t stopped the owners of private jets from objecting to the practice. Elon Musk notoriously threatened legal action and banned users from his “free speech” platform X for revealing the movements of his private jet, a practice he described as tantamount to sharing “assassination coordinates.” The use of luxury planes by public servants is a perennial political issue too. In January 2023, two years before he was named as his successor, Patel blasted FBI Director Christopher Wray’s use of the “tax payer funded private jet” based on exactly this same public tracking data.

Patel’s attitude has changed now that he enjoys free use of that jet. In October, Patel’s jet was monitored flying to to State College Regional Airport in Pennsylvania, a brief drive to an arena on the Penn State campus where he and girlfriend Alexis Wilkins were photographed at a Real American Freestyle pro-wrestling event at which Wilkins had performed a song. Flight data then showed the jet headed to Nashville later that evening after the wrestling match, where Wilkins lives, according to her personal website. After facing criticism for using a government plane to see his girlfriend sing the national anthem at a local wrestling event, Patel quickly lashed out in an X post, attacking public scrutiny of the flights and claiming the use of such data is “cowardly and jeopardizes our safety.”

Related

The Questionable Case of Kristi Noem’s $50 Million Luxury Jet

As FBI director, Patel is required by federal policy to use the jet for personal trips, but Patel is also required to reimburse the DOJ for personal flights. The FBI did not respond to a request for comment or questions posed by The Intercept, including whether Patel has reimbursed the DOJ for personal jet use.

Plane-tracking websites usually draw data from several sources, with many relying heavily on information straight from the FAA. Jet owners can ask the FAA to exclude their transponder data from public trackers, a request many commercial services will honor. For this reason, many plane-watching enthusiasts favor ADS-B Exchange, a free website that crowdsources transponder data collected by thousands of volunteers on the ground and pools it for public consumption. Because it uses crowdsourcing instead of just official FAA data, ADS-B Exchange shows every flight its connected antennae pick up — even if the aircraft’s owners have requested to be delisted. (Although it’s the most comprehensive, certain flights, like military planes that broadcast encrypted coordinates, can remain undetected even by ADS-B Exchange.)

Software engineer and plane tracking enthusiast John Wiseman recommended ADS-B Exchange and Airplanes.live, another service. “They don’t use FAA data, so they’re not bound by FAA rules on what data can be distributed. They also don’t take requests from aircraft owners to anonymize flights,” Wiseman explained.

ADS-B Exchange, with its sprawling, comprehensive map of nearly every plane in the sky, can be overwhelming at first. But tracking Patel’s jet (or any aircraft) is simple. Last year Congress made it more difficult for the public to connect private jets to their owners; establishing the ownership of non-governmental planes can sometimes require serious digging. But the existence of the FBI jet is a matter of public record, and its registration information is publicly available from the FAA’s website. Each plane in the FAA’s system has a unique number: Patel’s is N708JH. The FAA website confirms that N708JH is owned by an entity at 935 Pennsylvania Avenue in Washington, DC, the address of FBI headquarters. Searching ADS-B Exchange for N708JH will immediately pull up the plane’s current position, if in the air, as well as historical records of past flights. At the time of publication, the jet was last clocked landing at a municipal airport outside of Washington.

Pressing the “Play” button on ADS-B Exchange’s interface will animate a selected historical flight route. It’s possible to dive deeper into the data using filters; appending “?mil=1” at the end of the site’s URL will show only available military flights, for example.

Related

ICE Deportation Airline Avelo Relies on Blue-State Subsidies. Will Dem Governors Do Anything About It?

Anyone can search the FAA website to find other planes of interest; a search for the word “department” reveals dozens of aircraft registered to various federal offices. The website Planespotters.net collects runway photography from around the world, including of N708JH, and can be used to find other aircraft belonging to the Justice Department or other government entities. With a tail number in hand plucked either from FAA or another source, it’s not hard to figure out a plane’s location and travel history.

There are limits to what flight data can reveal. Nothing about the purpose of a flight is disclosed, nor are its passengers. But in conjunction with other open-source information — Wilkins’s Instagram account placed Patel alongside her at Real American Freestyle — it can help fill in the gaps.

The Wall Street Journal recently used flight data from the plane-tracking service FlightRadar24 in a recent report that showed not only had the FBI jet traveled to Nashville coinciding with Wilkins’s wrestling performance, but shortly thereafter also ferried Patel to the Boondoggle Ranch, a Texas hunting resort.

But if the U.S. government, particularly the military, wants to keep a flight hidden badly enough, it will. As the Pentagon launched a string of airstrikes against Iran in June, flight-tracking enthusiasts across the internet latched onto an Air Force refueling plane heading west, toward a base that houses B-2 bombers. Meanwhile, the actual B-2 bombers involved in the strike — which didn’t appear on any tracking portals — were flying in the opposite direction.

Savvy flight-watchers can sometimes get lucky, though, Wiseman explained. ADS-B Exchange also picks up planes broadcasting TIS-B signals, another transponder system. “Many law enforcement and military aircraft only show up on TIS-B. The icons are distinctive, and the [ID codes] are prefixed with ‘~’,” he said. “Sometimes people think those are drones, or fighter jets. They’re almost always just police aircraft, but once in a while, rarely, they are drones or fighter jets.”

Experts suggest new plane trackers chat with other enthusiasts who can share helpful knowledge and context. Discords or other online communities of plane-watchers can help newcomers avoid common errors, like mistaking what might be a typical flight pattern for something unusual or suspect.

Despite Patel’s characterization of plane scrutiny as the work of “clickbait haters” and “uninformed internet anarchists,” plane trackers who spoke to The Intercept all firmly defended the public’s right to know. “Public officials are acting in our name, so we should actively be making sure they’re doing what they claim to be doing and doing so ethically,” Canadian researcher Steffan Watkins, an avid tracker of military and other governmental flights, told The Intercept.

“There’s a strong public interest in knowing how aircraft are being used, and keeping government organizations accountable,” said Wiseman. “It’s good for the public, journalists, and researchers to be able to see how these aircraft are being used, in detail, and how public funds are being spent, in detail. Transparency also deters misuse.”

Wiseman recounted how flight tracking techniques have been used to reveal governmental aerial surveillance of protests: “If it hadn’t been for the persistence of a bunch of nerds obsessed with planes it’s possible the public would have never known.”

Even the DOJ itself has been a fan of ADS-B tracking. In 2016, Assistant Attorney General Leslie Caldwell said her office had used Dictator Alert, a plane-tracking website that uses ADS-B Exchange data, to aid in criminal seizure investigations.

Watkins rejected concerns — largely by the jet-owning class — that using ADS-B data presents a security risk: “Adversaries, like China, Russia, Iran, etc. already have better ways larger numbers of intelligence professionals tracking these movements, so the public should be as informed as public sources allow.” Wiseman agreed, saying, “There seems to be almost no risk to legitimate operations by making this information public, based on that fact that over the past years that ADS-B has been in wide use and registration information has been generally public we’ve seen very few if any incidents.”

The post How to Track Kash Patel’s Jet appeared first on The Intercept.

ICE is currently considering awarding contracts to companies interested in providing “skip tracing” services that would deploy privatize investigators to track down immigrants residing inside the U.S. The plan, first reported by The Intercept, states that these bounty hunters will be tasked with conducting surveillance and ultimately pinpointing the home address of “aliens,” defined by the Department of Homeland Security as “a person who is not a citizen or national of the United States.” They could earn bonus payments based on how many immigrants they help the government in apprehending, and how quickly.

In a letter sent Monday to DHS Secretary Kristi Noem, Rep. Krishnamoorthi states that the outsourcing proposal raises worrying questions about accountability.

“Allowing private contractors to perform enforcement activities under a system of performance-based financial incentives, essentially bounty hunting, outsources one of the government’s most coercive powers to actors who operate with little oversight and limited public accountability,” Krishnamoorthi writes. “These contractors are not subject to the same scrutiny, discipline, or transparency that restrains federal officers, and entrusting them with powers of investigation and surveillance risks creating an enforcement apparatus that functions beyond the reach of ordinary checks and balances. Once the state begins contracting out its power to police, it invites the very abuses, secrecy, and corruption our founders sought to prevent.”

Related

ICE Plans Cash Rewards for Private Bounty Hunters to Locate and Track Immigrants

Krishnamoorthi also notes his worries about how the plan will further blur the line between federal authority and the private sector, adding more corporate profit motive to the government’s increasingly chaotic and freewheeling domestic immigration operations. “In such a system built on quotas and cash rewards with minimal oversight, mistakes are not just possible — they are certain. The pressure to hit numbers replaces the judgment, training, and accountability that should define real law enforcement.”

In addition to registering his alarm over the proposal, Krishnamoorthi is asking Noem to clarify outstanding questions about how the private bounty hunter system would work, including whether contractors will be required to identify themselves as agents of the federal government.

ICE did not respond to questions about Krishnamoorthi’s inquiry, but said in a statement that “The Request for Information is solely for information and planning purposes and does not constitute a Request for Proposal nor does it restrict the Government to any acquisition approach. As part of its market research, ICE is issuing this RFI to determine the estimated number of interested vendors capable of meeting this requirement. The government may use the responses to this RFI for information and planning purposes.”

The post Lawmaker Challenges ICE Plan to Hire Bounty Hunters appeared first on The Intercept.

This summer, ICE renewed access to software tools for use by Homeland Security Investigations, or HSI, an enforcement division previously tasked with transnational crime that has become increasingly common on American streets under the Trump administration.

The $19,000 transaction, according to federal procurement data, provided “Nvidia software licenses, which will be used by Homeland Security Investigations to enhance data analysis & improve investigative capabilities through high-performance computing solutions.”

“HSI’s growing investment in LLMs” — large language models — “suggests that it may be investing in systems that can be used to surveil U.S. citizens, migrants, and visitors,” said Amos Toh, senior counsel at the Brennan Center for Justice.

Large language models can be used to draw inference by fusing people’s publicly available data, and might be used by ICE to “to identify persons of interest and generate investigative leads.” There are well-documented flaws, however, in the way the AI crunches data and reproduces biases.

Toh said, “These problems make it more likely that people will be targeted based on flawed intelligence.”

In a statement, ICE said, “Like other law enforcement agencies, ICE employs various forms of technology to investigate criminal activity and support law enforcement efforts while respecting civil liberties and privacy interests.”

When asked whether Nvidia had any ability to ensure ICE was using its technology lawfully, company spokesperson John Rizzo told The Intercept, “Millions of U.S. consumers, businesses, and government agencies use general-purpose computers every day. We do not and cannot monitor the use of general-purpose computers by U.S. government employees.”

Related

Mahmoud Khalil Won His Freedom Despite the Best Efforts of ICE’s Intelligence Unit

HSI’s mission has shifted during President Donald Trump’s second administration. The ICE division has long since assisted in civil immigration enforcement, but its focus was on criminal investigations such as drug smuggling and human trafficking.

“HSI has long sought to distance itself from ICE Enforcement and Removal Operations, which carries out basic immigration law enforcement,” said Aaron Reichlin-Melnick, a senior fellow at the American Immigration Council, told The Intercept. “On January 20, President Trump signed an executive order directing HSI to make immigration enforcement its top priority.”

Nvdia has been cozying up to Trump, who is threatening restrictions on chip exports to China, a lucrative market for the chipmaker. At a speech at a Nvidia tech conference in Washington on Tuesday, CEO Jensen Huang praised Trump and thanked those assembled “for your service and helping make America great again.”

How Nvidia Might Help ICE

How ICE plans to use Nvidia’s services is unclear; the specific software in question is not disclosed in the procurement documents.

Nvidia offers a variety of software-based services that could be useful for ICE data analysis. Nvidia has a dominant position across machine learning and artificial intelligence fields, including platforms to run large language models and video analytics.

The reseller through which ICE is buying access to Nvidia products, California-based New Tech Solutions, has previously sold the U.S. government licenses for “virtual workstations,” which essentially lease remote access to powerful chips known as graphics processing units, or GPUs, housed in data centers owned by Nvidia.

Such hardware could be used to train and query machine learning models. A 2023 report by the Department of Homeland Security on its potential usage of machine learning flagged HSI as standing to benefit from adopting the technology, including by rapidly searching and summarizing suspicious activity reports through large language models.

“HSI agents could quickly access and make sense of more than tens of millions of reports through ad hoc, unstructured queries over a voice interface,” the report says, adding that the system could also automatically scan and classify the contents of footage recorded by HSI agents.

A recently published inventory of ways DHS is using artificial intelligence tools reveals other areas where ICE may be able to make use of Nvidia’s “high-performance computing solutions.”

The document, which reflects Homeland Security practices as of July, notes HSI uses machine learning algorithms “to identify and extract critical evidence, relationships, and networks from mobile device data, leveraging machine learning capabilities to determine locations of interest.” The document also says HSI uses large language models to “identify the most relevant information in reports, accelerating investigative analysis by rapidly identifying persons of interest, surfacing trends, and detecting networks or fraud.”

HSI’s Shifting Mission

Procurement data about HIS’s use of Nvidia technology comes as ICE ramps up its presence in cities and towns across the U.S. Raids by ICE are viewed as being increasingly extreme and unchecked by legal or policy constraints, leading to aggressive protests against the immigrant enforcement.

HSI is playing a growing role in the controversial enforcement — and the crackdown on demonstrations.

Since Trump’s executive order on HSI, said Reichlin-Melnick of the American Immigration Council, “large numbers of agents have been reassigned away from criminal investigations to carry out immigration arrests instead.”

Related

Google Secretly Handed ICE Data About Pro-Palestine Student Activist

HSI agents in Washington have rounded up residents for minor traffic infractions and, earlier this month fired a gun into a man’s car. In June, HSI took part in the arrest of Newark Mayor Ras Baraka outside an ICE facility he was scheduled to tour with a delegation of New Jersey lawmakers. Charges of trespassing against Baraka were later dismissed.

HSI’s activities, though, go beyond street arrests and workplace raids: This week, 404 Media reported the agency was collecting utility customer data from Con Edison.

Like most large tech firms, Nvidia’s claims it adheres to various international human rights frameworks, including the Universal Declaration of Human Rights, which prohibits prejudice based on race or national origin.

The post ICE Investigations, Powered by Nvidia appeared first on The Intercept.

According to the document, which solicits information from interested contractors for a potentially forthcoming contract opportunity, companies hired by ICE will be given bundles of information on 10,000 immigrants at a time to locate, with further assignments provided in “increments of 10,000 up to 1,000,000.”

The solicitation says ICE is considering “monetary bonuses” paid out based on performance.

The solicitation says ICE is “exploring an incentive based pricing structure” to encourage quick results, with “monetary bonuses” paid out based on performance. For example, ICE says contractors might get paid a bonus for identifying a person’s correct address on the first try or finding 90 percent of its targets within a set timeframe. (ICE did not immediately respond to a request for comment.)

The document closely resembles a plan reportedly circulated by a group of military contractors, including former Blackwater CEO and Trump ally Erik Prince.

In February, Politico reported that Prince and other were pushing for the formation of a private efforts to locate immigrants and a “bounty program which provides a cash reward for each illegal alien held by a state or local law enforcement officer,” according pitch materials obtained by the outlet.

The proposal called for “skip-tracing,” a method of using available information to locate people — something ICE is already handing out multimillion-dollar contracts for, according to a recent report in The Lever.

Soon, according to the newly published procurement document, private sector ICE contractors will surveil and confirm the home or work addresses of tens of thousands of immigrants in the U.S. and then report their locations back to the government.

“DHS ICE has an immediate need for Skip Tracing and Process Serving Services using Government furnished case data with identifiable information, commercial data verification, and physical observation services, to verify alien address information, investigate alternative alien address information, confirm the new location of aliens, and deliver materials/documents to aliens as appropriate,” according to the October 31 request for information.

Data provided by ICE will include “case data” provided by the government, location data, social media information, as well as “photos and documents” showing where a person lives or works. With this in hand, contractors will surveil their target to confirm the accuracy of their home address, including “time-stamped photographs of the location,” before reporting back to ICE.

“The vendor should prioritize the alien’s residence,” the document notes, “but failing that will attempt to verify place of employment.”

The plan entails not just on-the-ground monitoring but the use of digital surveillance. ICE says contractors can use off-the-shelf surveillance technology to confirm immigrants’ addresses, including “Enhanced location research, which entails automated and manual real-time skip tracing.”

Surveillance tools that ingest and track mobile phone location data are widely available on the private market, many of them already used by ICE.

“Multiple verification sources are recommended to achieve a high confidence level,” the document says, encouraging potential vendors to use “all technology systems available.” 

The new procurement document notes that “the Government is contemplating awarding contracts to multiple vendors” due to the large number of immigrants whose whereabouts it seeks to confirm.

The post ICE Plans Cash Rewards for Private Bounty Hunters to Locate and Track Immigrants appeared first on The Intercept.

Amazon Web Services has furnished the Israeli government — including its military and intelligence agencies — with a suite of state-of-the-art data processing and storage services since 2021 as part of its controversial Project Nimbus deal. Last year, The Intercept revealed a provision in that contract requiring Amazon and Google, the other Nimbus vendor, to sell cloud services to Rafael Advanced Defense Systems and Israeli Aerospace Industries, two leading Israeli weapons firms.

New internal financial data and emails between Amazon personnel and their Israeli corporate and governmental clients show that Amazon has consistently provided software to both Rafael and IAI in 2024 and 2025 — periods during which Israel’s military was using their products to indiscriminately kill civilians and destroy civil infrastructure. Rafael purchased artificial intelligence technologies made available through Amazon Web Services, including the state-of-the-art large language model Claude, developed by AI startup Anthropic.

The materials reviewed by The Intercept also indicate Amazon sold cloud-computing services to Israel’s nuclear program and offices administering the West Bank, where Israeli military occupation, population displacement, and settlement construction is widely considered illegal under international law.

Amazon proclaims broad commitments to international human rights values, like most of its Big Tech peers. “We’re committed to identifying, assessing, prioritizing, and addressing adverse human rights impacts connected to our business,” the company’s Global Human Rights Principles website states. “Within Amazon’s own operations, we deploy a variety of mechanisms to conduct due diligence, assessing and responding to risks across the company,” including “human rights impact assessments to assess risks specific to Amazon businesses, including in the sectors and the countries where we operate.”

Amazon declined to comment or respond to a list of detailed questions, including whether it conducted a human rights impact assessment pertaining to selling its services to weapons companies whose products are used in a war widely assessed to be genocidal.

Rafael, Israel Aerospace Industries, and the Israeli Ministry of Defense did not respond to a request for comment.

It’s unclear how much money Rafael and IAI paid Amazon for its services. The documents reviewed by The Intercept show that Amazon sold its cloud-computing to Rafael at a discounted rate, though the exact percentage is not disclosed. The materials cite a 35 percent discount for services sold to the Israeli Ministry of Defense, a major Project Nimbus customer; it’s unclear if this rate is provided to Rafael and IAI as well.

Rafael was founded in 1948 as a governmental weapons research lab and, like its American equivalents at Raytheon or Lockheed, has become synonymous with Israeli militarism. Today, the state-owned company manufactures a diverse arsenal of missiles, bombs, drones, and other weaponry for both domestic use and international export. The corporation has thrived since Hamas’s October 7 attacks, reporting record revenues in both 2023 and 2024 that it attributed to Israel’s bombardment of Gaza. “2024 was a record year for Rafael, during the longest and most complex multi-front war in Israel’s history,” CEO Yoav Turgeman said last year, referring to the ongoing war with Hamas and related regional conflicts. “Rafael played a significant role in Israel’s military achievements in offense, intelligence and defense.”

IAI, another state-owned weapons firm, is best known for co-developing Israel’s anti-rocket Iron Dome system alongside Rafael. The company also manufactures a wide array of military aircraft, including its Heron line of drones — which the company has boasted about being used to great effect in Israel’s war on Gaza. A November 2023 promotional item about IAI’s drones published in the Jerusalem Post noted that “In the face of the October 7 challenges, the HERON UAS demonstrated its strategic importance by providing real-time intelligence, supporting targeted acquisitions, and aiding in the neutralization of threats.”

Missiles and other weapon systems built by Rafael and IAI have been used against Palestinians throughout the Gaza war. One of the most prominent Rafael weapons is its line of missile guidance kits dubbed SPICE: “Smart, Precise Impact, and Cost-Effective.” The SPICE technology converts “dumb” 1,000 or 2,000-pound bombs into “smart” guided munitions. In September 2024, Israel bombed a refugee camp — previously designated by the government as a “safe zone” for displaced Palestinians — with what weapons analysts later assessed was a 2,000 pound SPICE-guided bomb. The attack, condemned by the United Nations as “unconscionable,” killed at least 19 Palestinians, including women and children, with a massive explosion that burned, shredded, and in some cases buried those who’d sought shelter at the site. Fragments of a SPICE guidance kit were found amid the wreckage of a December 2024 airstrike on a house in Central Gaza that reportedly killed 12 civilians.

Retired Air Force operator and weapons targeting expert Wes Bryant described Rafael and IAI as “highly integral to Israel’s defense industrial complex,” telling The Intercept both companies are implicated in killing civilians. Israel has been criticized for its frequent use of 2,000-pound bombs in Gaza, one of the densest urban areas in the world. “It could level multiple large houses in the average suburban American neighborhood,” Bryant explained. “Ideally the only time they should be used in urban warfare is when we have identified a large and/or hardened enemy structure and confirmed it is entirely in use by the enemy and has no civilian function nor civilians within or around it at risk.”

Rafael’s electro-optically guided Spike family of missiles are designed to both punch through and destroy heavily armored tanks or kill humans, and can be fired from portable ground-launchers in addition to drones or other vehicles. Some Spike missiles use “shaped charge” warheads, which slice into targets with a cone of scalding metal launched from the weapon as it detonates. In 2009, a former Pentagon official described the Spike to Haaretz as “a special missile that is made to make very high-speed turns, so if you have a target that is moving and running away from you, you can chase him with the weapon.” Rafael marketing materials note one variant “can be used in urban combat against structural targets found in urban settings for in-structure detonation.” Arms experts have at times attributed devastating, widespread shrapnel wounds inflicted upon Palestinian civilians to Spike missiles, which can be packed with tiny pieces of tungsten. When a tungsten-loaded Spike weapon hits its target, the 3-millimeter metal cubes blast outward in a 65-foot radius, lacerating blood vessels, puncturing organs, and shredding the flesh of anyone nearby, according to analysts.

In April 2024, an investigation by The Times of London revealed Israel used a drone-launched Spike missile manufactured by Rafael to kill seven aid workers with World Central Kitchen. U.N. special rapporteur for the occupied Palestinian territories Francesca Albanese called for indictments following the attack, echoing international condemnations and demands for an inquiry into whether the airstrike constituted a war crime.

“Though the IDF does not release numbers of munitions utilized in the war in Gaza, SPIKE missiles have been used extensively and have been attributed by many investigations to the death of civilians, including children,” said Bryant. “It is likely that Israel has used dozens, if not hundreds, of SPIKE missiles throughout Gaza since the outset of the conflict.”

Both Rafael and IAI supply the Israeli military with so-called loitering munitions: suicide drones that can hover for extended periods while scanning for targets, then quickly slam into the ground and detonate an onboard explosive. Both companies’ weapons are frequently highlighted when the Israeli military–industrial apparatus wants to flag its technology supremacy. In July, Rafael posted a promotional video using footage of its Firefly suicide drone killing an apparently unarmed person walking down the street in an unidentified area of Gaza. Suicide drone attacks have also been documented in the Occupied West Bank; a December 2023 video captured a Firefly explosive descending into a dense courtyard.

Israel’s military similarly promoted the use of the shoulder-fired Matador rocket, co-developed by Rafael, in a March 2024 video reported by Israeli outlet Ynet: “In the clip, one of the terrorists opened fire from a room inside an apartment — and the use of a Matador missile targeting him precisely to eliminate the threat.” The outlet noted “a woman and two children” were in the adjoining room, but claimed they were not harmed in the missile attack against their home.

The Israeli military did not respond to a request for comment.

The documents show that Rafael acquired generative artificial intelligence tools through Amazon. In 2024, the firm sought to begin testing generative AI services made available through Amazon’s Bedrock service, which provides customers with machine-learning tools, including those made by third-party firms. According to the files, Rafael wanted to use both Amazon’s Titan G1 large language model and Claude, the advanced LLM model created by Anthropic.

Like its competitor OpenAI, Anthropic recently pivoted toward military contracting, announcing a $200 million deal with the Pentagon in July. Anthropic’s permissible use policy prohibits the use of its technology to “Produce, modify, design, or illegally acquire weapons,” and to “Design or develop weaponization and delivery processes for the deployment of weapons.” It’s unclear how the use of Claude by Rafael — a company that exists to design, develop, and deliver weapons — could be in compliance with this policy. The documents reviewed by The Intercept indicate Rafael was able to purchase access to these models, but do not reveal how they were used.

Anthropic did not respond to questions about Rafael’s usage of Claude, or whether it would permit a weapons company to use its services despite an apparent ban on exactly that. In a statement, spokesperson Eduardo Maia Silva said, “Anthropic services are available to users, including governments, in most countries and regions around the world under our standard commercial Usage Policy. Users are required to comply with our Usage Policies which include restrictions and prohibitions around how Claude can be deployed.”

Project Nimbus has been a military program from its start. The Israeli Ministry of Finance declared in 2021 that its purpose was “to provide the government, the defense establishment and others with an all encompassing cloud solution.” Google, Amazon’s co-contractor on the project, has repeatedly denied that Nimbus involves “highly sensitive, classified, or military workloads relevant to weapons or intelligence services,” while Amazon has generally refrained from commenting at all.

A separate internal Amazon document obtained by The Intercept shows that the company was quietly lobbying Israel to allow it to handle classified material from the country’s defense and intelligence community. The document, an overview of Israel’s regulatory landscape, explained that the country’s military and spy agencies were reluctant to migrate classified data onto Amazon’s cloud servers. But the paper also notes that Amazon was trying to influence state regulators into changing this position, and had begun working with one unnamed, major government body to bring some of its classified materials onto AWS.

Portions of the internal financial materials indicate exactly which Amazon services the Israeli military and state-owned weapons firms use. The purchases include dozens of networking, storage, and security tools, including Elastic Compute Cloud, which lets customers run software in virtual computers hosted by Amazon. Multiple documents show the Israeli Ministry of Defense purchased access to Amazon Rekognition, the company’s face-recognition tool, including an unspecified “OSINT,” or open-source intelligence, project by the Israeli military’s Central Command. Rekognition has previously been criticized for its lower accuracy rates with women and people of color; in 2020, the company announced a self-imposed yearlong moratorium on police use of Rekognition, citing the need for “stronger regulations to govern the ethical use of facial recognition technology.” The system, according to Amazon, is capable not only of identifying faces, but also a range of emotions including “fear.”

The documents show the Israeli military has also used Amazon technology to test large language models, though the specific models or applications are not mentioned. One Israeli military username includes the number 9900, a possible sign of use by the IDF’s Unit 9900, a geospatial intelligence unit that aided in planning strikes in Gaza, including through the use of a spy satellite developed by IAI. Unit 9900 also purchased cloud services from Microsoft, according to a January report by The Guardian and +972 Magazine.

The documents indicate that another Amazon customer through its Nimbus contract is the Israeli state-operated Soreq Nuclear Research Center, a scientific installation constructed in cooperation with the United States in the 1950s. Although Israel’s nuclear weapons arsenal is technically secret and unacknowledged by its government, Soreq operates in the open, ostensibly part of the country’s civilian atomic energy program. Unlike Israel’s highly classified Negev Nuclear Research Center, Soreq is not believed to be a major contributor to the country’s weapons capabilities. A 1987 Pentagon study, however, stated the Soreq installation “runs the full nuclear gamut of activities …required for nuclear weapons design and fabrication.” A 2002 report by the Stockholm International Peace Research Institute noted “The Soreq Center shares a security zone with the Palmikhim AB,” an Israeli Air Force base, “from where missiles are assembled and test launched into the Mediterranean Sea.”

A separate document briefly references as AWS users unspecified government offices in “Judea and Samaria,” Israel’s term for the West Bank, which it has illegally occupied since 1967. Ioannis Kalpouzos, a visiting professor at Harvard Law School and an expert on human rights law and laws of war, told The Intercept that Amazon’s work with Israeli weapons makers could potentially create liability under international law depending on “whether it is foreseeable that it will lead to the commission of international crimes.”

“There is no need for genocidal intent for accessorial liability in aiding the principal to commit genocide,” Kalpouzos said.

Related

Google Worried It Couldn’t Control How Israel Uses Project Nimbus, Files Reveal

It’s unclear to what extent Amazon is aware of how its services are being used by the companies that build Israel’s bombs or the military that drops them. The Intercept previously reported internal anxieties amid the bidding process at Google, where leadership fretted that the project was structured in such a way that the company would be kept in the dark about how exactly its technology would be used, potentially in violation of human rights standards. While servicing the Israeli government includes plenty of mundane applications — say transportation, schools, or hospitals — in addition to its military, there’s little nuance in the operations of Rafael and IAI. Even if Amazon lacks the ability to conduct oversight of these customers, Bryant said there is little ambiguity when it comes to the purpose of their business: building and selling weapons.

“I don’t see how Amazon can make a claim of not being complicit in killing,” said Bryant, who previously led civilian harm assessments at the Pentagon, “even if they don’t fully know what everything is used for.”

The post As Israel Bombed Gaza, Amazon Did Business With Its Bomb-Makers appeared first on The Intercept.

“We are the single worst technology to use to abuse civil liberties, which is by the way the reason why we could never get the NSA or the FBI to actually buy our product,” Karp said.

What he didn’t mention was the fact that a tranche of classified documents revealed by Edward Snowden and The Intercept in 2017 showed how Palantir software helped the National Security Agency and its allies spy on the entire planet.

Palantir has attracted increased scrutiny as the pace of its business with the federal government has surged during the second Trump administration. In May, the New York Times reported Palantir would play a central role in a White House plan to boost data sharing between federal agencies, “raising questions over whether he might compile a master list of personal information on Americans that could give him untold surveillance power.” Karp immediately rejected that report in a June interview on CNBC as “ridiculous shit,” adding that “if you wanted to use the deep state to unlawfully surveil people, the last platform on the world you would pick is Palantir.”

Karp made the same argument in this week’s podcast appearance, after “All-In” co-host David Sacks — the Trump administration AI and cryptocurrency czar — pressed him on matters of privacy, surveillance, and civil liberties. “One of the criticisms or concerns that I hear on the right or from civil libertarians is that Palantir has a large-scale data collection program on American citizens,” Sacks said.

Karp replied by alleging that he had been approached by a Democratic presidential administration and asked to build a database of Muslims. “We’ve never done anything like this. I’ve never done anything like this,” Karp said, arguing that safeguards built into Palantir would make it undesirable for signals intelligence. That’s when he said the company’s refusal to abuse civil liberties is “the reason why we could never get the NSA or the FBI to actually buy our product.”

Karp later stated: “To your questions, no, we are not surveilling,” taking a beat before adding, “uh, U.S. citizens.”

Related

How Peter Thiel’s Palantir Helped the NSA Spy on the Whole World

In 2017, The Intercept published documents originally provided by Snowden, a whistleblower and former NSA contractor, demonstrating how Palantir software was used in conjunction with a signals intelligence tool codenamed XKEYSCORE, one of the most explosive revelations from the NSA whistleblower’s 2013 disclosures. XKEYSCORE provided the NSA and its foreign partners with a means of easily searching through immense troves of data and metadata covertly siphoned across the entire global internet, from emails and Facebook messages to webcam footage and web browsing. A 2008 NSA presentation describes how XKEYSCORE could be used to detect “Someone whose language is out of place for the region they are in,” “Someone who is using encryption,” or “Someone searching the web for suspicious stuff.”

Later in 2017, BuzzFeed News reported Palantir’s working relationship with the NSA had ceased two years prior, citing an internal presentation delivered by Karp. Palantir did not provide comment for either The Intercept’s or BuzzFeed News’ reporting on its NSA work.

The Snowden documents describe how intelligence data queried through XKEYSCORE could be imported straight into Palantir software for further analysis. One document mentions use of Palantir tools in “Mastering The Internet,” a joint NSA/GCHQ mass surveillance initiative that included pulling data directly from the global fiber optic cable network that underpins the internet. References inside HTML files from the NSA’s Intellipedia, an in-house reference index, included multiple nods to the company, such as “Palantir Classification Helper,” “[Target Knowledge Base] to Palantir PXML,” and “PalantirAuthService.”

And although Karp scoffed at the idea that Palantir software would be suitable for “deep state” usage, a British intelligence document note also published by The Intercept quotes GCHQ saying the company’s tools were developed “through [an] iterative collaboration between Palantir computer scientists and analysts from various intelligence agencies over the course of nearly three years.”

Karp’s carefully worded clarification that Palantir doesn’t participate in the surveillance of Americans specifically would have been difficult if not impossible for the company to establish with any certainty. From the moment of its disclosure, XKEYSCORE presented immense privacy and civil liberties threats, both to Americans and noncitizens alike. But in the United States, much of the debate centered around the question of how much data on U.S. citizens is ingested — intentionally or otherwise — by the NSA’s globe-spanning surveillance capabilities.

Even without the NSA directly targeting Americans, their online speech and other activity is swept up during the the agency’s efforts to spy on foreigners: say, if a U.S. citizen were to email a noncitizen who is later targeted by the agency. Even if the public takes the NSA at its word that it does not deliberately collect and process information on Americans through tools like XKEYSCORE, it claims the legal authority under Section 702 of the Foreign Intelligence Surveillance Act to subsequently share such data it “incidentally” collects with other U.S. agencies, including the FBI.

The legality of such collection remains contested. Legal loopholes created in the name of counterterrorism and national security leave large gaps through which the NSA and its partner agencies can effectively bypass legal protections against spying on Americans and the 4th Amendment’s guarantee against warrantless searches.

A 2014 report by The Guardian on the collection of webcam footage explained that GCHQ, the U.K.’s equivalent of the NSA, “does not have the technical means to make sure no images of UK or US citizens are collected and stored by the system, and there are no restrictions under UK law to prevent Americans’ images being accessed by British analysts without an individual warrant.” The report notes “Webcam information was fed into NSA’s XKeyscore search tool.”

In 2021, the federal Privacy and Civil Liberties Oversight Board concluded a five-year investigation into XKEYSCORE. In declassified remarks reported by the Washington Post, Travis LeBlanc, a board member who took part in the inquiry, said the NSA’s analysis justifying XKEYSCORE’s legality “lacks any consideration of recent relevant Fourth Amendment case law on electronic surveillance that one would expect to be considered.”

“The former Board majority failed to ask critical questions like how much the program costs financially to operate, how many U.S. persons have been impacted by KEYSCORE,” his statement continued. “While inadvertently or incidentally intercepted communications of U.S. persons is a casualty of modern signals intelligence, the mere inadvertent or incidental collection of those communications does not strip affected U.S. persons of their constitutional or other legal rights.”

Palantir did not respond when asked by The Intercept about the discrepancy between its CEO’s public remarks and its documented history helping spy agencies at home and abroad use what the NSA once described as its “widest reaching” tool.

The post Alex Karp Insists Palantir Doesn’t Spy on Americans. Here’s What He’s Not Saying. appeared first on The Intercept.

The document, a sort of special operations wishlist of near-future military technology, reveals new details about a broad variety of capabilities that SOCOM hopes to purchase within the next five to seven years, including state-of-the-art cameras, sensors, directed energy weapons, and other gadgets to help operators find and kill their quarry. Among the tech it wants to procure is machine-learning software that can be used for information warfare.

To bolster its “Advanced Technology Augmentations to Military Information Support Operations” — also known as MISO — SOCOM is looking for a contractor that can “Provide a capability leveraging agentic Al or multi‐LLM agent systems with specialized roles to increase the scale of influence operations.”

So-called “agentic” systems use machine-learning models purported to operate with minimal human instruction or oversight. These systems can be used in conjunction with large language models, or LLMs, like ChatGPT, which generate text based on user prompts. While much marketing hype orbits around these agentic systems and LLMs for their potential to execute mundane tasks like online shopping and booking tickets, SOCOM believes the techniques could be well suited for running an autonomous propaganda outfit.

“The information environment moves too fast for military remembers [sic] to adequately engage and influence an audience on the internet,” the document notes. “Having a program built to support our objectives can enable us to control narratives and influence audiences in real time.”

Laws and Pentagon policy generally prohibit military propaganda campaigns from targeting U.S. audiences, but the porous nature of the internet makes that difficult to ensure.

In a statement, SOCOM spokesperson Dan Lessard acknowledged that SOCOM is pursuing “cutting-edge, AI-enabled capabilities.”

“All AI-enabled capabilities are developed and employed under the Department of Defense’s Responsible AI framework, which ensures accountability and transparency by requiring human oversight and decision-making,” he told The Intercept. “USSOCOM’s internet-based MISO efforts are aligned with U.S. law and policy. These operations do not target the American public and are designed to support national security objectives in the face of increasingly complex global challenges.”

Tools like OpenAI’s ChatGPT or Google’s Gemini have surged in popularity despite their propensity for factual errors and other erratic outputs. But their ability to immediately churn out text on virtually any subject, written in virtually any tone — from casual trolling to pseudo-academic — could mark a major leap forward for internet propagandists. These tools give users the potential to finetune messaging any number of audiences without the time or cost of human labor.

Whether AI-generated propaganda works remains an open question, but the practice has already been amply documented in the wild. In May 2024, OpenAI issued a report revealing efforts by Iranian, Chinese, and Russian actors to use the company’s tools to engage in covert influence campaigns, but found none had been particularly successful. In comments before the 2023 Senate AI Insight Forum, Jessica Brandt of the Brookings Institution warned “LLMs could increase the personalization, and therefore the persuasiveness, of information campaigns.” In an online ecosystem filled with AI information warfare campaigns, “skepticism about the existence of objective truth is likely to increase,” she cautioned. A 2024 study published in the academic journal PNAS Nexus found that “language models can generate text that is nearly as persuasive for US audiences as content we sourced from real-world foreign covert propaganda campaigns.”

Related

OpenAI’s Pitch to Trump: Rank the World on U.S. Tech Interests

Unsurprisingly, the national security establishment is now insisting that the threat posed by this technology in the hands of foreign powers, namely Russia and China, is most dire.

“The Era of A.I. Propaganda Has Arrived, and America Must Act,” warned a recent New York Times opinion essay on GoLaxy, software created by the Chinese firm Beijing Thinker originally used to play the board game Go. Co-authors Brett Benson, a political science professor at Vanderbilt University, and Brett Goldstein, a former Department of Defense official, paint a grim picture showing GoLaxy as an emerging leader in state-aligned influence campaigns.

GoLaxy, they caution, is able to scan public social media content and produce bespoke propaganda campaigns. “The company privately claims that it can use a new technology to reshape and influence public opinion on behalf of the Chinese government,” according to a companion piece by Times national security reporter Julian Barnes headlined “China Turns to A.I. in Information Warfare.” The news item strikes a similarly stark tone: “GoLaxy can quickly craft responses that reinforce the Chinese government’s views and counter opposing arguments. Once put into use, such posts could drown out organic debate with propaganda.” According to these materials, the Times says, GoLaxy has “undertaken influence campaigns in Hong Kong and Taiwan, and collected data on members of Congress and other influential Americans.”

To respond to this foreign threat, Benson and Goldstein argue a “coordinated response” across government, academia, and the private sector is necessary. They describe this response as defensive in nature: mapping and countering foreign AI propaganda.

That’s not what the document from the Special Operations Forces Acquisition, Technology, and Logistics Center suggests the Pentagon is seeking.

The material shows SOCOM believes it needs technology that closely matches the reported Chinese capabilities, with bots scouring and ingesting large volumes of internet chatter to better persuade a targeted population, or an individual, on any given subject.

SOCOM says it specifically wants “automated systems to scrape the information environment, analyze the situation and respond with messages that are in line with MISO objectives. This technology should be able to respond to post(s), suppress dissenting arguments, and produce source material that can be referenced to support friendly arguments and messages.”

The Pentagon is paying especially close attention to those who might call out its propaganda efforts.

“This program should also be able to access profiles, networks, and systems of individuals or groups that are attempting to counter or discredit our messages,” the document notes. “The capability should utilize information gained to create a more targeted message to influence that specific individual or group.”

SOCOM anticipates using generative systems to both craft propaganda messaging and simulate how this propaganda will be received once sent into the wild, the document notes. SOCOM hopes it will use “agentic systems that replicate specific knowledge, skills, abilities, personality traits, and sociocultural attributes required for different roles of individuals comprising a team,” before moving on to “brainstorm and test operational campaigns against agent‐based replicas of individuals and groups.” These simulations are more elaborate than focus groups, calling instead for “comprehensive models of entire societies to enable MISO planners to use these models to experiment or test various multiple scenarios.”

The SOCOM wishlist continues to include a need for offensive deepfake capabilities, first reported by The Intercept in 2023.

The prospect of LLMs creating an infinite firehose of expertly crafted propaganda has been received by alarm — but generally in the context of the United States as target, not perpetrator.

A 2023 publication by the State Department-funded nonprofit Freedom House warned of “The Repressive Power of Artificial Intelligence,” predicting “AI-assisted disinformation campaigns will skyrocket as malicious actors develop additional ways to bypass safeguards and exploit open-source models.” Warning that “Generative AI draws authoritarian attention,” the Freedom House report cites potential use by China and Russia, but only mentions domestic use of the technology in a brief section about the presidential campaigns of Ron DeSantis and Donald Trump, as well as a deepfake video of Joe Biden manipulated to depict the former president making transphobic comments. The extent to which an automated propaganda machine capable of global reach warrants public concern depends on the scope of its application, according to Andrew Lohn, former director for emerging technology on the National Security Council.

“I would not be so concerned if some foreign soldiers are wrongly convinced that our special operation is going to happen Wednesday morning by helicopter from the east rather than Tuesday night by boat from the west,” said Lohn, now a senior fellow at Georgetown’s Center for Security and Emerging Technology.

The military has a history of manipulating civilian populations for political or ideological purposes. A troubling example was uncovered in 2024, when Reuters reported the Defense Department had operated a clandestine anti-vax social media campaign to undercut public confidence in the Chinese Covid vaccine, fearing its efficacy might draw Asian countries closer to a major geopolitical rival. Pentagon-created tweets described the Chinese Sinovac-CoronaVac shot — described by the World Health Organization as “safe and effective” — as “fake” and untrustworthy. According to the Reuters report, then-Special Operations Command Pacific General Jonathan Braga “pressed his bosses in Washington to fight back in the so-called information space” by backing the clandestine propaganda campaign.

William Marcellino, a behavioral scientist at the RAND Corporation focusing on the geopolitics of machine-learning systems and Pentagon procurement, told The Intercept such systems are being built out of necessity. “Regimes like those from China and Russia are engaged in AI-enabled, at-scale malign influence efforts,” he said. State-affiliated groups in China, he warned, “have explicitly designed AI at-scale systems for public opinion warfare.”

“Countering those campaigns likely requires AI at-scale responses,” he said.

SOCOM has in recent years been public about its desire for AI-created propaganda systems. These statements suggest a broader interest that includes influence operations against entire populations, as opposed to narrowly tailored toward military personnel.

In 2019, a senior Pentagon special operations official spoke at a defense symposium of the country’s “need to move beyond our 20th century approach to messaging and start looking at influence as an integral aspect of modern irregular warfare.” The official noted that this “will also require new partnerships beyond traditional actors, throughout the world, through efforts to amplify voices of [non-governmental organizations] and individual citizens who bring transparency to malign activities of our competitors.” The following year, then-SOCOM commander Gen. Richard Clarke described his interest in using AI to achieve these ends.

“As we look at the ability to influence and shape in this [information] environment, we’re going to have to have artificial intelligence and machine learning tools,” Clarke said in 2020 remarks first reported by National Defense Magazine, “specifically for information ops that hit a very broad portfolio, because we’re going to have to understand how the adversary is thinking, how the population is thinking, and work in these spaces.”

Heidy Khlaaf, chief scientist at the AI Now Institute and former safety engineer at OpenAI, warned against a fighting-fire-with-fire approach: “Framing the use of generative and agentic AI as merely a mitigation to adversaries’ use is a misrepresentation of this technology, as offensive and defensive uses are really two sides of the same coin and would allow them to use it precisely in the same way that adversaries do.”

Automated online influence campaigns might wind up having lackluster results, according to Emerson Brooking, a senior fellow at the Atlantic Council’s Digital Forensic Research Lab. “Russia has been using AI programs to automate its influence operations. The program is not very good,” he said.

The tendency of LLMs to fabricate falsehoods and perpetuate preconceptions when prompted by users could also prove a major liability, Brooking warned. “Tasked with figuring out the ‘hearts and minds’ of a complex and understudied country, they may lean heavily on an AI to help them, which will be likely to tell them what they already want to hear,” he said.

Khlaaf added that “agentic” systems, heavily marketed by tech firms as independent digital brains, are still error-prone and unpredictable. “The introduction of agentic AI in these disinformation campaigns adds a layer of both safety and security concerns, as several research results have demonstrated how easily we can compromise and divert the behavior of agentic AI,” she told The Intercept. “With these security issues unresolved, [SOCOM] risks that their campaigns are not only compromised, but that they produce material that was not intended.”

Brooking, who previously worked as an adviser to the Office of the Under Secretary of Defense for Policy on cybersecurity matters, also pointed to the mixed track record of prior U.S. online propaganda efforts. In 2022, researchers revealed a network of Twitter and Facebook accounts secretly operated by U.S. Central Command that had been pushing bogus news articles containing anti-Russian and Iranian talking points. The network, which failed to gain traction on either social network, quickly became an embarrassment for the Pentagon.

“We know from other public reporting that the U.S. has long sought to ‘suppress dissenting arguments’ and generate positive press in certain areas of operation,” he said. “We also know that these efforts have not worked very well and can be deeply embarrassing or counterproductive when revealed to the American public. AI tends to make these campaigns stupider, not more effective.”

The post Pentagon Document: U.S. Wants to “Suppress Dissenting Arguments” Using AI Propaganda appeared first on The Intercept.

Johnny “Taz” Mulford belongs to a Florida chapter of the Infidels, a biker group for veterans of U.S. wars and private military contractors like Blackwater. In May, Mulford began recruiting among his Facebook network for an unspecified job opportunity, asking anyone who “can still shoot, move and communicate” to contact him.

Reached by phone on Friday, Mulford confirmed to The Intercept that he is currently in Israel, adding that he was “on his way to a checkpoint,” but declined to comment further. Two sources directly familiar with the Gaza operations of UG Solutions, including former contractor Anthony Aguilar, confirmed Mulford’s employment to The Intercept. Mulford’s ties to the motorcycle group were first reported by Zeteo.

UG Solutions is a contractor providing security at aid distribution sites run by GHF, the aid effort in Gaza backed by the Trump administration and Israel.

“They’re in a primary Arab Muslim population, delivering food at the end of the gun.”

Mulford’s membership in the Infidels and numerous tattoos widely linked to the Crusades and contemporary far-right movements raise questions about his role as a contractor for the GHF mission. Among other posts on Facebook, Mulford nods to Christian Zionism by sharing a post calling Israel “God’s chosen nation” and a video mocking pro-Palestine protesters.

“If I went into Israel with a Nazi swastika on my arm and said ‘Heil Hitler,’ what would people think of me?” said Aguilar, a former Green Beret and UG Solutions contractor who has become a public critic of the GHF, raising concerns about Mulford’s tattoos and Infidels affiliation in the Middle East. “They’re in a primary Arab Muslim population, delivering food at the end of the gun.”

Mulford and the GHF did not immediately respond to requests for comment. The Infidels national umbrella and a local Florida chapter did not respond to requests made, respectively, to an online form and a chapter official.

“Johnny Mulford is a respected contractor in the industry, with over 30 years of cleared service supporting U.S. government and allied efforts. Any allegations suggesting otherwise are categorically false and defamatory,” Drew O’Brien, a UG Solutions spokesperson, said in a statement. “We do not screen for personal hobbies or affiliations unrelated to job performance or security standards. Every team member undergoes comprehensive background checks, and only qualified, vetted individuals are deployed on UG Solutions operations.”

O’Brien declined to comment on Mulford’s tattoos.

The Infidels were founded in 2006 by an American mercenary in Iraq nicknamed “Slingshot,” according to the club’s website, which says the early members were security contractors and military veterans. According to its website, “Infidels Motorcycle Club is a veteran formed and based MC for Patriotic Americans and our supporting allies.”

“Bearing in mind that we support the war against terrorism, and many of our Club members have and are serving in Iraq and other locations worldwide as either members of the military or as civilian contractors, our political views may not be shared by everyone,” says the national umbrella group in a Facebook post. “We neither support nor tolerate the Jihadist movement and those who support it. If on the other hand you do support the country’s efforts against Islamic extremism, then support your local Infidels MC!”

In 2015, the Infidels’ Colorado Springs chapter threw a pig roast barbecue party “in defiance of the Muslim holiday of Ramadan,” according to an event flyer that also “included comparisons of Muslim men to pedophiles,” a local outlet reported at the time.

Mulford, who registered the local chapter in Florida, is an active member of the Infidels, according to his and others’ social media postings. He is frequently shown in photos online posing with fellow club members at meet-ups. The Infidels wear matching leather vests bearing the club name and a red cross on the back. In one photo, Mulford’s vest has an embroidered patch on the front that says “Original Infidel.”

Crusader Iconography

The Infidels — including Mulford — frequently employ Crusader iconography in their tattoos and apparel.

Photos of Mulford show him tattooed with crosses affiliated with the Crusades and, more recently, right-wing Christian movements.

A photo of Mulford on Facebook shows him without a shirt after an apparent outing to fish. On his right forearm is an American flag rendered in flames and overlain by a so-called Templar symbol: a shield emblazoned with a red cross, styled after the Christian military order of the Knights Templar. His left bicep displays another Templar shield. A tattoo on his right forearm displays the Jerusalem or Crusader cross, a squared-off cross with smaller crosses in each of the corners.

In another photo, Mulford can be seen wearing a vest that includes both the Infidels name and an amalgamation of several Crusader-style crosses.

According to Matthew Gabriele, a medieval studies professor at Virginia Tech and an expert in Crusader iconography, the Jerusalem cross and Templar shield are frequently embraced by white supremacists and the far right, — a nod to an imagined “existential conflict between Islam and Christianity” in the Middle Age, Gabriele said. Crusader iconography of this kind doesn’t reflect the historical record, but rather a sort of Christian revenge fantasy.

“It doesn’t have a whole lot of specific attachment the Middle Ages themselves, but a nostalgic version in which this existential conflict between Islam and Christianity, that has gone back to Islam’s founding, has always put Christianity on the defensive,” Gabriele said. Crosses and shields “symbolize that during the Crusades, Christianity struck back in a positive way. It really is a particular stance toward Islam and the Middle East.”

The Crusader aesthetic and the proud self-labeling of oneself as an “infidel” grew in popularity during the war on terror and have remained as gestures of anti-Muslim sentiment on the right.

“It was a way for a particular kind of American soldier,” Gabriele said, “to kind of reflect back Al Qaeda’s rhetoric: ‘Yeah we are the crusaders, we’re going to come there and kick your ass.’”

Crusader symbols have attracted scrutiny when worn by figures like Defense Secretary Pete Hegseth, who was questioned about his Jerusalem cross tattoo during Senate confirmation hearings. Hegseth defended the cross as a symbol of Christianity.

In July, while the GHF’s food distributions were ongoing, Mulford posted an illustration on Facebook of a kneeling Crusader knight with a glowing cross in the background and a superimposed biblical quote.

Other Facebook photos shared by Mulford show him with Crusader-style crosses on his arms and the number 1095 across his chest — the year Pope Urban II called for the First Crusade.

The date of 1095 has been cited as symbolically important by violent right-wing actors, Gabriele said, from Norwegian mass shooter Anders Breivik to Brenton Tarrant, perpetrator of the anti-Muslim massacre in Christchurch, New Zealand.

The date 1095, Gabriele said, represents a worldview in which Muslims are “a threat to be killed” and driven from the Holy Land.

In 2018, the national Infidels umbrella group shared a photo montage from what it describes as a “Crusader ride” organized by its members.

Security Contractor

Mulford served in the Marine Corps from 1982 to 1985 before a stint in the Army from 1987 to 2007, when he saw multiple deployments to Iraq and Afghanistan, according to an Army spokesperson. Subsequently, according to a personal website, he served overseas as a “security agent” for a “government agency.”

His service records include numerous commendations and achievements. A report stated that Mulford was “debarred” following “nonjudicial punishment” under military disciplinary rules related to an alleged scheme where “Mulford received a kickback from a contractor who provided skydiving training to Fort Bragg Soldiers.” Mulford’s debarment, a designation the military uses in deciding to award contracts, was terminated in 2007, shortly before his retirement from the Army at the rank of master sergeant.

The company employing Mulford is one of at least three U.S. contractors for the GHF, the nonprofit distributing food in Gaza with Israel’s approval.

To distribute what it says are over 108 million meals so far, the foundation has hired a pair of U.S. companies — one helmed by a former CIA official, the other by a Green Beret veteran — to provide logistics and armed private security contractors. Other aid organizations say the idea of staffing aid distribution sites with armed contractors violates basic principles of neutrality and have refused to work with the GHF.

One of UG Solutions’ partner organizations has already drawn scrutiny for its leader’s views on Islam and Palestinians.

In July, independent journalist Jack Poulson reported that Matthew Murphy, the president of a small relief organization called the Sentinel Foundation that partnered with UG Solutions to distribute aid in Gaza earlier this year, had a record of making bigoted remarks against Muslims generally and Palestinians in particular. In a podcast interview last year, Murphy referred to Palestine as “a little shithole.”

“Killing and beheading and raping and treating, you know, Christian and Jewish women as lesser-than and slaves is not just something terrorists think, it’s Islam,” Murphy said.

The Sentinel Foundation was co-founded by former Green Beret Jameson Govoni, who went on to found UG Solutions.

The GHF and its partners have drawn worldwide scrutiny since they began aid distributions in May.

At least 1,373 Palestinians have been killed seeking food since the foundation began its work in Gaza, including 859 people near distribution sites and 514 along food convoy routes, according to the United Nations. Palestinians say that many have died under gunfire from the Israeli military.

In job listings, UG Solutions describes itself as “a fast-moving, mission-driven private security company with global reach.” The Charlotte-based company first got involved in the conflict earlier this year when its private soldiers were tasked with manning checkpoints during a ceasefire.

The company has sought out former U.S. Special Forces veterans, according to job listings.

Four Democratic members of Congress last week wrote to UG Solutions and another GHF contractor, warning them that the companies’ employees could be held liable if war crimes have been committed. Working closely with the Israeli military, those members warned, has exposed the company’s staffers to great legal risk.

UG Solutions has denied mistreating Palestinians in Gaza, while acknowledging that its contractors have used pepper spray and “warning shots” to disperse crowds.

The post Team Leader at Gaza Aid Distribution Sites Belongs to Anti-“Jihad” Motorcycle Club, Has Crusader Tattoos appeared first on The Intercept.

A CBP presentation for an “Industry Day” summit with private sector vendors, obtained by The Intercept, lays out a detailed wish list of tech CBP hopes to purchase, like satellite connectivity for surveillance towers along the border and improved radio communications. But it also shows that state-of-the-art, AI-augmented surveillance technologies will be central to the Trump administration’s anti-immigrant campaign, which will extend deep into the interior of the North American continent, hundreds of miles from international borders as commonly understood.

Related

Google Is Helping the Trump Administration Deploy AI Along the Mexican Border

The recent passage of Trump’s sprawling flagship legislation funnels tens of billions of dollars to the Department of Homeland Security. While much of that funding will go to Immigration and Customs Enforcement to bolster the administration’s arrest and deportation operations, a great deal is earmarked to purchase new technology and equipment for federal offices tasked with preventing immigrants from arriving in the first place: Customs and Border Protection, which administers the country’s border surveillance apparatus, and its subsidiary, the U.S. Border Patrol.

One page of the presentation, describing the wishlist of Border Patrol’s Law Enforcement Operations Division, says the agency needs “Advanced AI to identify and track suspicious activity in urban environment [sic],” citing the “challenges” posed by “Dense residential areas.” What’s considered “suspicious activity” is left unmentioned.

Customs and Border Protection did not respond to questions posed about the slides by The Intercept.

The reference to AI-aided urban surveillance appears on a page dedicated to the operational needs of Border Patrol’s “Coastal AOR,” or area of responsibility, encompassing the entire southeast of the United States, from Kentucky to Florida. A page describing the “Southern AOR,” which includes all of inland Nevada and Oklahoma, similarly states the need for “Advanced intelligence to identify suspicious patterns” and “Long-range surveillance” because “city environments make it difficult to separate normal activity from suspicious activity.”

Related

Crossing the U.S. Border? Here’s How to Protect Yourself

Although the Fourth Amendment provides protection against arbitrary police searches, federal law grants immigration agencies the power to conduct warrantless detentions and searches within 100 miles of the land borders with Canada, Mexico, or the coastline of the United States. This zone includes most of the largest cities in the United States, including Los Angeles, New York, as well as the entirety of Florida.

The document mentions no specific surveillance methods or “advanced AI” tools that might be used in urban environments. Across the Southwest, residents of towns like Nogales and Calexico are already subjected to monitoring from surveillance towers placed in their neighborhoods. A 2014 DHS border surveillance privacy impact assessment warned these towers “may capture information about individuals or activities that are beyond the scope of CBP’s authorities. Video cameras can capture individuals entering places or engaging in activities as they relate to their daily lives because the border includes populated areas,” for example, “video of an individual entering a doctor’s office, attending public rallies, social events or meetings, or associating with other individuals.”

Last year, the Government Accountability Office found the DHS tower surveillance program failed six out of six privacy policies designed to prevent such overreach. CBP is also already known to use “artificial intelligence” tools to ferret out “suspicious activity,” according to agency documents. A 2024 inventory of DHS AI applications includes the Rapid Tactical Operations Reconnaissance program, or RAPTOR, which “leverages Artificial Intelligence (AI) to enhance border security through real-time surveillance and reconnaissance. The AI system processes data from radar, infrared sensors, and video surveillance to detect and track suspicious activities along U.S. borders.”

The document’s call for urban surveillance reflect the reality of Border Patrol, an agency empowered, despite its name, with broad legal authority to operate throughout the United States.

“Border Patrol’s escalating immigration raids and protest crackdowns show us the agency operates heavily in cities, not just remote deserts,” said Spencer Reynolds, a former attorney with the Department of Homeland Security who focused on intelligence matters. “Day by day, its activities appear less based on suspicion and more reliant on racial and ethnic profiling. References to operations in ‘dense residential areas’ are alarming in that they potentially signal planning for expanded operations or tracking in American neighborhoods.”

Automating immigration enforcement has been a Homeland Security priority for years, as exemplified by the bipartisan push to expand the use of machine learning-based surveillance towers like those sold by arms-maker Anduril Industries across the southern border. “Autonomous technologies will improve the USBP’s ability to detect, identify, and classify potential threats in the operating environment,” according to the agency’s 2024 – 2028 strategy document. “After a threat has been identified and classified, autonomous technology will enable the USBP to track threats in near real-time through an integrated network.”

The automation desired by Border Patrol seems to lean heavily on computer vision, a form of machine learning that excels at pattern matching to find objects in the desert that resemble people, cars, or other “items of interest,” rather than requiring crews of human agents to monitor camera feeds and other sensors around the clock. The Border Patrol presentation includes multiple requests for small drones that incorporate artificial intelligence technologies to aid in the “detection, tracking, and classification” of targets.

A computer system that has analyzed a large number of photographs of trucks driving through the desert can become effective at identifying similar vehicles in the future. But efforts to algorithmically label human behavior as “suspicious” — an abstract concept compared to “truck” — based only on its appearance has been criticized by some artificial intelligence scholars and civil libertarians as error-prone, overly subjective if not outright pseudoscientific, and often reliant on ethnic and religious stereotypes. Any effort to apply predictive techniques based on surveillance data from entire urban areas or residential communities would exacerbate these risks of bias and inaccuracy.

“In the best of times, oversight of technology and data at DHS is weak and has allowed profiling, but in recent months the administration has intentionally further undermined DHS accountability,” explained Reynolds, now senior counsel at the Brennan Center’s liberty and national security program. “Artificial intelligence development is opaque, even more so when it relies on private contractors that are unaccountable to the public — like those Border Patrol wants to hire. Injecting AI into an environment full of biased data and black-box intelligence systems will likely only increase risk and further embolden the agency’s increasingly aggressive behavior.”

The desire to hunt “suspicious” people with “advanced AI” reflects a longtime ambition at the Department of Homeland Security, Mohammad Tajsar, an attorney at the ACLU of Southern California, told The Intercept. Military and intelligence agencies across the world are increasingly working to use forms of machine learning, often large language models like OpenAI’s GPT, to rapidly ingest and analyze varied data sources to find buried trends, threats, and targets — though systemic issues with accuracy remain unsolved.

This proposed use case dovetails perfectly with the Homeland Security ethos, Tajsar said. “They’re addicted to suspicious activity reporting because they fundamentally believe that their targets do suspicious things, and that suspicious things can predict criminal behavior,” a notion Tajsar described as a “fantasy” that “remains unchallenged despite the complete lack of empiricism to support it.” With the rapid proliferation of technologies billed as artificially intelligent, “they think that they can bring to bear all of their disparate sources of data using computers, and they see that as a breakthrough in what they’ve been trying to do for a long, long time.”

While much of the presentation addresses Border Patrol’s wide-ranging surveillance agenda, it also includes information about other departmental tech needs.

The Border Patrol Tactical Unit, or BORTAC, exists on paper to execute domestic missions involving terrorism, hostage situations, or other high-risk scenarios. But the unit has become increasingly associated with suppressing dissent and routine deportation raids: In 2020, the Trump administration ordered BORTAC into the streets of Portland to tamp down protests, and the special operations unit has been similarly deployed in Los Angeles this year.

According to the presentation, CBP hopes to arm the already heavily militarized BORTAC with the ability to see through walls in order to “detect people within a structure or rubble.”

Another page of the document, listing the agency’s “Subterranean Portfolio,” claims CBP is preparing to lay an additional 2,100 miles of fiber optic cable along the northern and southern border in order to detect passing migrants, as part of a sensor network that also includes seismic, laser, visual, and cellular tracking.

The post Border Patrol Wants Advanced AI to Spy on American Cities appeared first on The Intercept.

The legislation, signed into law by President Donald Trump on July 4, provides significant spending increases to military and law enforcement projects, including over $6 billion for various border security technologies. Among these initiatives is expanding the ever-widening “virtual wall” of sensor-laden surveillance towers along the U.S.-Mexico border, where computers increasingly carry out the work of detecting and apprehending migrants.

Related

Google Is Helping the Trump Administration Deploy AI Along the Mexican Border

Anduril, today a full-fledged military contractor, got its start selling software-augmented surveillance towers to CBP. Anduril has pitched its Sentry Tower line on the strength of its “autonomous” capabilities, which use machine learning software to perpetually scan the horizon for possible objects of interest — i.e. people attempting to cross the border — rather than requiring a human to monitor sensor feeds. Thanks to bipartisan support for the vision of a border locked down by computerized eyes, Anduril has become a dominant player in border surveillance, edging out incumbents like Elbit and General Dynamics.

Now, that position looks to be enshrined in law: A provision buried in the new mega-legislation stipulates that none of the $6 billion border tech payday can be spent on border towers unless they’ve been “tested and accepted by U.S. Customs and Border Protection to deliver autonomous capabilities.”

The bill defines “autonomous” as “a system designed to apply artificial intelligence, machine learning, computer vision, or other algorithms to accurately detect, identify, classify, and track items of interest in real time such that the system can make operational adjustments without the active engagement of personnel or continuous human command or control.”

Anduril is now the country’s only approved border tower vendor.

That reads like a description of Anduril’s product — because it might as well be. A CBP spokesperson confirmed to The Intercept that under the new law, Anduril is now the country’s only approved border tower vendor. Although CBP’s plans for border surveillance tend to be in flux, Homeland Security presentation documents have cited the need for hundreds of new towers in the near future, money that for the time being will only be available to Anduril.

Anduril did not respond to a request for comment.

In a statement to The Intercept, Dave Maass, investigations director at the Electronic Frontier Foundation and a longtime observer of border surveillance technologies, raised concerns about the apparent codification of militarized AI at the border. “I was cynically expecting Trump’s bill to quadruple-down on wasteful surveillance technology at the border, but I was not expecting language that appears to grant an exclusive license to Anduril to install AI-powered towers,” he said. “For 25 years, surveillance towers have enriched influential contractors while delivering little security, and it appears this pattern isn’t going to change anytime soon. Will this mean more towers in public parks and AI monitoring the everyday affairs of border neighborhoods? Most likely. Taxpayers will continue footing the bill while border communities will pay an additional price with their privacy and human rights.”

The law’s stipulation is not only a major boon to Anduril, but a blow to its competitors, now essentially locked out of a lucrative and burgeoning market until they can gain the same certification from CBP. In April, The Intercept reported on a project to add machine learning surveillance capabilities to older towers manufactured by the Israeli military contractor Elbit, and General Dynamics’ Information Technology division has spent years implementing an upgraded version of its Remote Video Surveillance System towers. The fate of these companies’ border business is now unclear.

Elbit did not respond to a request for comment. General Dynamics Information Technology spokesperson Jay Srinivasan did not respond directly when asked about Anduril’s border exclusivity deal, stating, “We can’t speculate on the government’s acquisition strategy and the different contracts it intends to use to exercise the funding,” and pointed to the company’s existing surveillance tower contracts.

Towers like Anduril’s Sentry have proven controversial, hailed by advocates in Silicon Valley and Capitol Hill as a cheaper and more humane way of stopping illegal immigration than building a physical wall, but derided by critics as both ineffective and invasive. Although border towers are frequently marketed with imagery of a lone edifice in a barren desert, CBP has erected surveillance towers, which claim to offer detailed, 24/7 visibility for miles, in populated residential areas. In 2024, the Government Accountability Office reported CBP’s surveillance tower program failed to address all six of the main privacy protections that were supposed to be in place, including a rule that “DHS should collect only PII [Personally Identifiable Information] that is directly relevant and necessary to accomplish the specified purpose(s).”

The new Trump spending package emphasizes the development and purchase of additional autonomous and unmanned military hardware that could prove favorable for Anduril, which has developed a suite of military products that run on machine learning-centric software. One section, for instance, sets aside $1.3 billion for “for expansion of unmanned underwater vehicle production,” an initiative that could dovetail with Anduril’s announcement last year that it would open a 100,000–150,000 square foot facility in Rhode Island dedicated to building autonomous underwater vehicles. Another sets aside $200 million for “the development, procurement, and integration of mass-producible autonomous underwater munitions,” which could describe Anduril’s Copperhead line of self-driving torpedoes, announced in April.

The bill also earmarks billions for suicide attack drones and counter-drone weaponry, technologies also sold by Anduril. Anduril is by no means the only contractor who can provide this weaponry, but it already has billions of dollars worth of contracts with the Pentagon for similar products, and enjoys a particularly friendly relationship with the Trump administration.

Trae Stephens, Anduril’s co-founder and executive chairman, served on Trump’s transition team in 2016 and was reportedly floated for a senior Pentagon position late last year. Michael Obadal, Trump’s nominee for Under Secretary of the Army, worked at Anduril until June, according to his Linkedin, and has come under fire for his refusal to divest his Anduril stock.

Anduril founder Palmer Luckey is also longtime Trump supporter, and has hosted multiple fundraisers for his presidential campaigns.

Following Trump’s reelection last fall, Luckey told CNBC he was sanguine about his company’s fortunes in the new administration. “We did well under Trump, and we did better under Biden,” he said of Anduril. “I think we will do even better now.”

Update: July 9, 2025, 4:45 p.m. ET
This story has been updated to specify that CBP is contracting with General Dynamics’ Information Technology business unit for its border tower program.

The post Trump’s Big Beautiful Gift to Anduril appeared first on The Intercept.

The meteoric machine-learning firm announced itself to the world in a December 2015 press release that lays out a vision of technology to benefit all people as people, not citizens. There are neither good guys nor adversaries. “Our goal is to advance digital intelligence in the way that is most likely to benefit humanity as a whole,” the announcement stated with confidence. “Since our research is free from financial obligations, we can better focus on a positive human impact.”

Early rhetoric from the company and its CEO, Sam Altman, described advanced artificial intelligence as a harbinger of a globalist utopia, a technology that wouldn’t be walled off by national or corporate boundaries but enjoyed together by the species that birthed it. In an early interview with Altman and fellow OpenAI co-founder Elon Musk, Altman described a vision of artificial intelligence “freely owned by the world” in common. When Vanity Fair asked in a 2015 interview why the company hadn’t set out as a for-profit venture, Altman replied: “I think that the misaligned incentives there would be suboptimal to the world as a whole.”

Times have changed. And OpenAI wants the White House to think it has too.

In a March 13 white paper submitted directly to the Trump administration, OpenAI’s global affairs chief Chris Lehane pitched a near future of AI built for the explicit purpose of maintaining American hegemony and thwarting the interests of its geopolitical competitors — specifically China. The policy paper’s mentions of freedom abound, but the proposal’s true byword is national security.

OpenAI never attempts to reconcile its full-throated support of American security with its claims to work for the whole planet, not a single country. After opening with a quotation from Trump’s own executive order on AI, the action plan proposes that the government create a direct line for the AI industry to reach the entire national security community, work with OpenAI “to develop custom models for national security,” and increase intelligence sharing between industry and spy agencies “to mitigate national security risks,” namely from China.

In the place of techno-globalism, OpenAI outlines a Cold Warrior exhortation to divide the world into camps. OpenAI will ally with those “countries who prefer to build AI on democratic rails,” and get them to commit to “deploy AI in line with democratic principles set out by the US government.”

The rhetoric seems pulled directly from the keyboard of an “America First” foreign policy hawk like Marco Rubio or Rep. Mike Gallagher, not a company whose website still endorses the goal of lifting up the whole world. The word “humanity,” in fact, never appears in the action plan.

Rather, the plan asks Trump, to whom Altman donated $1 million for his inauguration ceremony, to “ensure that American-led AI prevails over CCP-led AI” — the Chinese Communist Party — “securing both American leadership on AI and a brighter future for all Americans.”

It’s an inherently nationalist pitch: The concepts of “democratic values” and “democratic infrastructure” are both left largely undefined beyond their American-ness. What is democratic AI? American AI. What is American AI? The AI of freedom. And regulation of any kind, of course, “may hinder our economic competitiveness and undermine our national security,” Lehane writes, suggesting a total merging of corporate and national interests.

Related

Trump’s Big, Beautiful Handout to the AI Industry

In an emailed statement, OpenAI spokesperson Liz Bourgeois declined to explain the company’s nationalist pivot but defended its national security work.

“We believe working closely with the U.S. government is critical to advancing our mission of ensuring AGI benefits all of humanity,” Bourgeois wrote. “The U.S. is uniquely positioned to help shape global norms around safe, secure, and broadly beneficial AI development—rooted in democratic values and international collaboration.”

The Intercept is currently suing OpenAI in federal court over the company’s use of copyrighted articles to train its chatbot ChatGPT.

OpenAI’s newfound patriotism is loud. But is it real? 

In his 2015 interview with Musk, Altman spoke of artificial intelligence as a technology so special and so powerful that it ought to transcend national considerations. Pressed on OpenAI’s goal to share artificial intelligence technology globally rather than keeping it under domestic control, Altman provided an answer far more ambivalent than the company’s current day mega-patriotism: “If only one person gets to have it, how do you decide if that should be Google or the U.S. government or the Chinese government or ISIS or who?” 

He also said, in the early days of OpenAI, that there may be limits to what his company might do for his country.

“I unabashedly love this country, which is the greatest country in the world,” Altman told the New Yorker in 2016. “But some things we will never do with the Department of Defense.” In the profile, he expressed ambivalence about overtures to OpenAI from then-Secretary of Defense Ashton Carter, who envisioned using the company’s tools for targeting purposes. At the time, this would have run afoul of the company’s own ethical guidelines, which for years stated explicitly that customers could not use its services for “military and warfare” purposes, writing off any Pentagon contracting entirely. 

Related

OpenAI Quietly Deletes Ban on Using ChatGPT for “Military and Warfare”

In January 2024, The Intercept reported that OpenAI had deleted this military contracting ban from its policies without explanation or announcement. Asked about how the policy reversal might affect business with other countries in an interview with Bloomberg, OpenAI executive Anna Makanju said the company is “focused on United States national security agencies.” But insiders who spoke with The Intercept on conditions of anonymity suggested that the company’s turn to jingoism may come more from opportunism than patriotism. Though Altman has long been on the record as endorsing corporate support of the United States, under an administration where the personal favor of the president means far more than the will of lawmakers, parroting muscular foreign policy rhetoric is good for business.

One OpenAI source who spoke with The Intercept recalled concerned discussions about the possibility that the U.S. government would nationalize the company. They said that at times, this was discussed with the company’s head of national security partnerships, Katrina Mulligan. Mulligan joined the company in February 2024 after a career in the U.S. intelligence and military establishment, including leading the media and public policy response to Edward Snowden’s leaks while on the Obama National Security Council staff, working for the director of national intelligence, serving as a senior civilian overseeing Special Operations forces in the Pentagon, and working as chief of staff to the secretary of the Army.

This source speculated that fostering closeness with the government was one method of fending off the potential risk of nationalization.

As an independent research organization with ostensibly noble, global goals, OpenAI may have been less equipped to beat back regulatory intervention, a second former OpenAI employee suggested. What we see now, they said, is the company “transitioning from presenting themselves as a nonprofit with very altruistic, pro-humanity aims, to presenting themselves as an economic and military powerhouse that the government needs to support, shelter, and cut red tape on behalf of.”

The second source said they believed the national security rhetoric was indicative of OpenAI “sucking up to the administration,” not a genuinely held commitment by executives.

“In terms of how decisions were actually made, what seemed to be the deciding factor was basically how can OpenAI win the race rather than anything to do with either humanity or national security,” they added. “In today’s political environment, it’s a winning move with the administration to talk about America winning and national security and stuff like that. But you should not confuse that for the actual thing that’s driving decision-making internally.”

The person said that talk of preventing Chinese dominance over artificial intelligence likely reflects business, not political, anxieties. “I think that’s not their goal,” they said. “I think their goal is to maintain their own control over the most powerful stuff.” 

But even if its motivations are cynical, company sources told The Intercept that national security considerations still pervaded OpenAI. The first source recalled a member of OpenAI’s corporate security team regularly engaging with the U.S. intelligence community to safeguard the company’s ultra-valuable machine-learning models. The second recalled concern about the extent of the government’s relationship — and potential control over — OpenAI’s technology. A common fear among AI safety researchers is a future scenario in which artificial intelligence models begin autonomously designing newer versions, ad infinitum, leading human engineers to lose control.

“One reason why the military AI angle could be bad for safety is that you end up getting the same sort of thing with AIs designing successors designing successors, except that it’s happening in a military black project instead of in a somewhat more transparent corporation,” the second source said. 

“Occasionally there’d be talk of, like, eventually the government will wake up, and there’ll be a nuclear power plant next to a data center next to a bunker, and we’ll all be moved into the bunker so that we can, like, beat China by managing an intelligence explosion,” they added. At a company that recruits top engineering talent internationally, the prospect of American dominance of a technology they believe could be cataclysmic was at times disquieting. “I remember I also talked to some people who work at OpenAI who weren’t from the U.S. who were feeling kind of sad about that and being like, ‘What’s going to happen to my country after the U.S. gets all the super intelligences?’”

Sincerity aside, OpenAI has spent the past year training its corporate algorithm on flag-waving, defense lobbying, and a strident anticommunism that smacks more of the John Birch Society than the Whole Earth Catalog.

In his white paper, Lehane, a former press secretary for Vice President Al Gore and special counsel to President Bill Clinton, advocates not for a globalist techno-utopia in which artificial intelligence jointly benefits the world, but a benevolent jingoism in which freedom and prosperity is underwritten by the guarantee of American dominance. While the document notes fleetingly, in its very last line, the idea of “work toward AI that benefits everyone,” the pitch is not one of true global benefit, but of American prosperity that trickles down to its allies.

Related

Why an “AI Race” Between the U.S. and China Is a Terrible, Terrible Idea

The company proposes strict rules walling off parts of the world, namely China, from AI’s benefits, on the grounds that they are simply too dangerous to be trusted. OpenAI explicitly advocates for conceiving of the AI market not as an international one, but “the entire world less the PRC” — the People’s Republic of China — “and its few allies,” a line that quietly excludes over 1 billion people from the humanity the company says it wishes to benefit and millions who live under U.S.-allied authoritarian rule. 

In pursuit of “democratic values,” OpenAI proposes dividing the entire planet into three tiers. At the top: “Countries that commit to democratic AI principles by deploying AI systems in ways that promote more freedoms for their citizens could be considered Tier I countries.” Given the earlier mention of building “AI in line with democratic principles set out by the US government,” this group’s membership is clear: the United States, and its friends.

Beneath them are Tier 2 countries, a geopolitical purgatory defined only as those that have failed to sufficiently enforce American export control policies and protect American intellectual property from Tier 3: Communist China. “CCP-led China, along with a small cohort of countries aligned with the CCP, would represent its own category that is prohibited from accessing democratic AI systems,” the paper explains. To keep these barriers intact — while allowing for the chance that Tier 2 countries might someday graduate to the top — OpenAI suggests coordinating “global bans on CCP-aligned AI” and “prohibiting relationships” between other countries and China’s military or intelligence services.

One of the former OpenAI employees said concern about China at times circulated throughout the company. “Definitely concerns about espionage came up,” this source said, “including ‘Are particular people who work at the company spies or agents?’” At one point, they said, a colleague worried about a specific co-worker they’d learned was the child of a Chinese government official. The sourced recalled “some people being very upset about the implication” that the company had been infiltrated by foreigners, while others wanted an actual answer: “‘Is anyone who works at the company a spy or foreign agent?’”

The company’s public adoration of Western democracy is not without wrinkles. In early May, OpenAI announced an initiative to build data centers and customized ChatGPT bots with foreign governments, as part of its $500 billion “Project Stargate” AI infrastructure construction blitz.

“This is a moment when we need to act to support countries around the world that would prefer to build on democratic AI rails, and provide a clear alternative to authoritarian versions of AI that would deploy it to consolidate power,” the announcement read.

Unmentioned in that celebration of AI democracy is the fact that Project Stargate’s financial backers include the government of Abu Dhabi, an absolute monarchy. On May 23, Altman tweeted that it was “great to work with the UAE” on Stargate, describing co-investor and Emirati national security adviser Tahnoun bin Zayed Al Nahyan as a “great supporter of openai, a true believer in AGI, and a dear personal friend.” In 2019, Reuters revealed how a team of mercenary hackers working for Emirati intelligence under Tahnoun had illegally broken into the devices of targets around the world, including American citizens.

Asked how a close partnership with an authoritarian Emirati autocracy fit into its broader mission of spreading democratic values, OpenAI pointed to a recent op-ed in The Hill in which Lehane discusses the partnership.

“We’re working closely with American officials to ensure our international partnerships meet the highest standards of security and compliance,” Lehane writes, adding, “Authoritarian regimes would be excluded.”

OpenAI’s new direction has been reflected in its hiring.

Since hiring Mulligan, the company has continued to expand its D.C. operation. Mulligan works on national security policy with a team of former Department of Defense, NSA, CIA, and Special Operations personnel. Gabrielle Tarini joined the company after almost two years at the Defense Department, where she worked on “Indo-Pacific security affairs” and “China policy,” according to LinkedIn. Sasha Baker, who runs national security policy, joined after years at the National Security Council and Pentagon.

The list goes on: Other policy team hires at OpenAI include veterans of the NSA, a Pentagon former special operations and South China Sea expert, and a graduate of the CIA’s Sherman Kent School for Intelligence Analysis. OpenAI’s military and intelligence revolving door continues to turn: At the end of April, the company recruited Alexis Bonnell, the former chief information officer of the Air Force Research Laboratory. Recent job openings have included a “Relationship Manager” focusing on “strategic relationships with U.S. government customers.”

Mulligan, the head of national security policy and partnerships, is both deeply connected to the defense and intelligence apparatus, and adept at the kind of ethically ambivalent thinking common to the tech sector.

“Not everything that has happened at Guantanamo Bay is to be praised, that’s for sure, but [Khalid Sheikh Mohammed] admitting to his crimes, even all these years later, is a big moment for many (including me),” she posted last year. In a March podcast appearance, Mulligan noted she worked on “Gitmo rendition, detention, and interrogation” during her time in government.

Mulligan’s public rhetoric matches the ideological drift of a company that today seems more concerned with “competition” and “adversaries” than kumbaya globalism. 

On LinkedIn, she seems to embody the contradiction between a global mission and full-throated alignment with American policy values. “I’m excited to be joining OpenAI to help them ensure that AI is safe and beneficial to all of humanity,” she wrote upon her hiring from the Pentagon.

Since then, she has regularly represented OpenAI’s interests and American interests as one and the same, sharing national security truisms such as “In a competition with China, the pace of AI adoption matters,” or “The United States’ continued lead on AI is essential to our national security and economic competitiveness,” or “Congress needs to make some decisive investments to ensure the U.S. national security community has the resources to harness the advantage the U.S. has on this technology.” 

Related

Trump’s Pick for a Top Army Job Works at a Weapons Company — And Won’t Give Up His Stock

This is to some extent conventional wisdom of the country’s past 100 years: A strong, powerful America is good for the whole world. But OpenAI has shifted from an organization that believed its tech would lift up the whole world, unbounded by national borders, to one that talks like Lockheed Martin. Part of OpenAI’s national security realignment has come in the form of occasional “disruption” reports detailing how the company detected and neutralized “malicious use” of its tools by foreign governments, coincidentally almost all of them considered adversaries of the United States. 

As the provider of services like ChatGPT, OpenAI has near-total visibility into how the tools are used or misused by individuals, what the company describes in one report as its “unique vantage point.” The reports detail not only how these governments attempted to use ChatGPT, but also the steps OpenAI took to thwart them, described by the company as an “effort to support broader efforts by U.S. and allied governments.” Each report has focused almost entirely on malign AI uses by “state affiliated” actors from Iran, China, North Korea, and Russia. A May 2024 report outed an Israeli propaganda effort using ChatGPT but stopped short of connecting it to that country’s government.

Earlier this month, representatives of the intelligence agency and the contractors who serve them gathered at the America’s Center Convention Complex in St. Louis for the GEOINT Symposium, dedicated to geospatial intelligence, the form of tradecraft analyzing satellite and other imagery of the planet to achieve military and intelligence objectives. 

On May 20, Mulligan took to the stage to demonstrate how OpenAI’s services could help U.S. spy agencies and the Pentagon better exploit the Earth’s surface. Though the government’s practice of GEOINT frequently ends in the act of killing, Mulligan used a gentler example, demonstrating the ability of ChatGPT to pinpoint the location where a photograph of a rabbit was taken. It was nothing if not a sales pitch, one predicated on the fear that some other country might leap at the opportunity before the United States.

“Government often feels like using AI is too risky and that it’s better and safer to keep doing things the way that we’ve always done them, and I think this is the most dangerous mix of all,” Mulligan told her audience. “If we keep doing things the way that we always have, and our adversaries adapt to this technology before we do, they will have all of the advantages that I show you today, and we will not be safer.” 

The post OpenAI’s Pitch to Trump: Rank the World on U.S. Tech Interests appeared first on The Intercept.

According to internal communications reviewed by The Intercept, employees on Wednesday began noticing that email messages sent from their company account containing a handful of keywords related to Palestine and Israel’s ongoing war in Gaza were not transmitted as expected. In some cases, employees say the emails arrived after many hours. Other emails never even made it to the intended recipient’s inbox at all.

Keywords subject to the disruption, according to employee test messages shared with The Intercept, include “Palestine,” “Gaza,” “apartheid,” and “genocide.” The word “Palestinian” does not appear affected, nor did emails containing deliberate misspellings of the word “Palestine.” Emails mentioning Israel appear to have gone through immediately.

The outage was first reported by The Verge.

In an email to The Intercept, Microsoft spokesperson Frank Shaw confirmed and defended the blockage. “Emailing large numbers of employees about any topic not related to work is not appropriate. We have a established forum for employees who have opted in to political issues. Over the past couple of days, a number of politically focused emails have been sent to tens of thousands of employees across the company and we have taken measures to try and reduce those emails to those that have not opted in.”

The heavy-handed approach, however, is not just deterring messages sent to large numbers of recipients, but also blocking all emails mentioning Palestine.

Following an April 7 protest at an event celebrating Microsoft’s 50th anniversary, two employees “sent separate emails to thousands of coworkers, calling on Microsoft to cut its contracts with the Israeli government,” The Verge reported.

Related

The Microsoft Police State: Mass Surveillance, Facial Recognition, and the Azure Cloud

The email disruption comes after multiple demonstrations at the four-day Microsoft Build developer conference this week. The protests were organized by current and former Microsoft employees with No Azure for Apartheid, an advocacy group demanding the suspension of the company’s work with the Israeli government.

In February, The Associated Press reported usage of Microsoft’s Azure cloud computing services by the Israeli military “skyrocketed” at the start of its ongoing bombardment of Gaza, which has now killed over 53,000 Palestinians. Earlier this month, the company absolved itself of wrongdoing in Gaza following an unspecified internal and external review. While Microsoft claimed “we have found no evidence that Microsoft’s Azure and AI technologies, or any of our other software, have been used to harm people,” the company also noted, “It is important to acknowledge that Microsoft does not have visibility into how customers use our software on their own servers or other devices.”

The post Microsoft Says It’s Censoring Employee Emails Containing the Word “Palestine” appeared first on The Intercept.

So the government has a plan for a one-stop shop.

The Office of the Director of National Intelligence is working on a system to centralize and “streamline” the use of commercially available information, or CAI, like location data derived from mobile ads, by American spy agencies, according to contract documents reviewed by The Intercept. The data portal will include information deemed by the ODNI as highly sensitive, that which can be “misused to cause substantial harm, embarrassment, and inconvenience to U.S. persons.” The documents state spy agencies will use the web portal not just to search through reams of private data, but also run them through artificial intelligence tools for further analysis.

Rather than each agency purchasing CAI individually, as has been the case until now, the “Intelligence Community Data Consortium” will provide a single convenient web-based storefront for searching and accessing this data, along with a “data marketplace” for purchasing “the best data at the best price,” faster than ever before, according to the documents. It will be designed for the 18 different federal agencies and offices that make up the U.S. intelligence community, including the National Security Agency, CIA, FBI Intelligence Branch, and Homeland Security’s Office of Intelligence and Analysis — though one document suggests the portal will also be used by agencies not directly related to intelligence or defense.

“In practice, the Data Consortium would provide a one-stop shop for agencies to cheaply purchase access to vast amounts of Americans’ sensitive information from commercial entities, sidestepping constitutional and statutory privacy protections,” said Emile Ayoub, a lawyer with the Brennan Center’s liberty and national security program.

“ODNI is working to streamline a number of inefficient processes, including duplicative contracts to access existing data, and ensuring Americans civil liberties and Fourth Amendment rights are upheld,” ODNI spokesperson Olivia Coleman said in a statement to The Intercept. Coleman did not answer when asked if the new platform would sell access to data on U.S. citizens, or how it would make use of artificial intelligence.

Related

IRS, Department of Homeland Security Contracted Firm That Sells Location Data Harvested From Dating Apps

Spy agencies and military intelligence offices have for years freely purchased sensitive personal information rather than obtain it by dint of a judge’s sign-off. Thanks largely to unscrupulous advertisers and app-makers working in a regulatory vacuum, it’s trivial to procure extremely sensitive information about virtually anyone with an online presence. Smartphones in particular leave behind immense plumes of data, including detailed records of your movement that can be bought and sold by anyone with an interest. The ODNI has previously defined “sensitive” CAI as information “not widely known about an individual that could be used to cause harm to the person’s reputation, emotional well-being, or physical safety.” Procurement documents reviewed by The Intercept make clear the project is designed to provide access to this highest “sensitive” tier of CAI.

The documents provide a glimpse at some of the many types of CAI available, including “information addressing economic security, supply chain, critical infrastructure protection, great power competition, agricultural data, industrial data, sentiment analysis, and video analytic services.”

While the proliferation of data that can reveal intimate details about virtually anyone has alarmed civil libertarians, privacy advocates, and certain members of Congress, the intelligence community sees another problem: There’s too much data to keep organized, and the disorganized process of buying it is wasting money. To address this overabundance, the ODNI is seeking private sector vendors to build and manage a new “commercial data consortium that unifies commercial data acquisition then enables IC users to access and interact with this commercial data in one place,” according to one procurement document obtained by The Intercept.

The ODNI says the platform, the “Intelligence Community (IC) Data Consortium (ICDC),” will help correct the currently “fragmented and decentralized” purchase of commercial data like smartphone location pings, real estate records, biometric data, and social media content. The document laments how often various spy agencies are buying the same data without realizing it. The ODNI says this new platform, which will live at www.icdata.gov, will “help streamline access to CAI for the entire IC and make it available to mission users in a more cohesive, efficient, and cost-effective manner by avoiding duplicative purchases, preventing sunk costs from unused licenses, and reducing overall data storage and compute costs,” while also incorporating “civil liberties and privacy best practices.”

While the project’s nod to civil liberties might come as some relief to privacy advocates, the project also represents the extent to which the use of this inherently controversial form of surveillance is here to stay. “Clearly the IC is still adhering to the ‘just grab all of it, we’ll find something to do with it’ mentality rather than being remotely thoughtful about only collecting data it needs or has a specific envisioned use for,” said Calli Schroeder, senior counsel at the Electronic Privacy Information Project.

Once the website is up and running, the procurement materials say the portal will eventually allow users to analyze the data using large language models, AI-based text tools prone to major factual errors and fabrications. The portal will also facilitate “sentiment analysis,” an often pseudoscientific endeavor purporting to discern one’s opinion about a given topic using implicit signals in their behavior, movement, or speech.

Such analysis is a “huge cause for concern” according to Schroeder. “It means the intelligence community is still, to at least some degree, buying into the false promise of a constantly and continuously debunked practice,” she said. “Let me be clear: Sentiment analysis not only does not work, it cannot work. Its only consistent success has been in perpetuating harmful discrimination (of gender, culture, race, and neurodivergence, among others).”

Whether for sentiment analysis or some other goal, using CAI data sets to query an AI crystal ball poses serious risks, said Ayoub. If such analysis worked as billed, “AI tools make it easier to extract, re-identify, and infer sensitive information about people’s identities, locations, ideologies, and habits — amplifying risks to Americans’ privacy and freedoms of speech and association,” he said. On top of that, “These tools are a black box with little insight into training data, metric, or reliability of outcomes. The IC’s use of these tools typically comes with high risk, questionable track records, and little accountability, especially now that AI policy safeguards were rescinded early in this administration.”

In 2023, the ODNI declassified a 37-page report detailing the vastly expanding use of such CAI data by the U.S. intelligence community, and the threat this poses to the millions of Americans whose lives are cataloged, packaged, and sold by a galaxy of unregulated data brokers. The report, drafted for then-director of national intelligence Avril Haines, included a dire warning to the public: “Today, in a way that far fewer Americans seem to understand, and even fewer of them can avoid, CAI includes information on nearly everyone that is of a type and level of sensitivity that historically could have been obtained, if at all, only through targeted (and predicated) collection, and that could be used to cause harm to an individual’s reputation, emotional well-being, or physical safety.”

Related

American Phone-Tracking Firm Demo’d Surveillance Powers by Spying on CIA and NSA

The extent to which CAI has commodified spy powers previously attainable only by well-resourced governments cannot be overstated: In 2021, for instance, The Intercept reported the existence of Anomaly Six, a startup that buys geolocational data leaked from smartphones apps. During an Anomaly Six presentation, the company demonstrated its ability to track not only the Chinese navy through the phones of its sailors, but also follow CIA and NSA employees as they commuted to and from work.

The ICDC project reflects a fundamental dissonance within the intelligence community, which acknowledges that CAI is a major threat to the public while refusing to cease buying it. “The government would never have been permitted to compel billions of people to carry location tracking devices on their persons at all times, to log and track most of their social interactions, or to keep flawless records of all their reading habits,” the ODNI wrote in its 2022 report. While conceding “unfettered access to CAI increases its power in ways that may exceed our constitutional traditions or other societal expectations,” the report says, “the IC cannot willingly blind itself to this information.”

In 2024, following the declassified report and the alarm it generated, the ODNI put forth a set of CAI usage rules purporting to establish guardrails against privacy violations and other abuses. The framework earned praise from some corners for requiring the intelligence community to assess the origin and sensitivity of CAI before using it, and for placing more rigorous requirements on agencies that wish to use the most intimate forms of private data. But critics were quick to point out that the ODNI’s rules, which enshrined the intelligence community’s “flexibility to experiment” with CAI, amounted to more self-regulation from a part of the government with a poor track record of self-regulating.

While sensitive CAI comes with more rules — like keeping records of its use, protecting its storage, and some disclosure requirements — these guidelines offer great deal latitude to the intelligence community. The rule about creating a paper trail pertaining to sensitive CAI use, for example, is mandated only “to the extent practicable and consistent with the need to protect intelligence sources and methods,” and can be ignored entirely in “exigent circumstances.” In other words, it’s not really a requirement at all.

Ayoub told The Intercept he worries the ICDC plan will only entrench this self-policing approach. The documents note that vendors would be tasked to some extent with determining whether the data they sell is indeed sensitive, and therefore subject to stricter privacy safeguards, rather than a third party. “Relying on private vendors to determine whether CAI is considered sensitive may increase the risk that the IC purchases known categories of sensitive information without sufficient safeguards for privacy and civil liberties or the warrant, court order, or subpoena they would otherwise need to obtain,” he said.

The portal idea appears to have started under the Biden administration, when it was known as the “Data Co-Op.” It now looks like it will go live during a Trump administration. Elon Musk’s so-called Department of Government Efficiency is already working on building and streamlining access to other large repositories of perilously sensitive information. In March, the Washington Post reported that DOGE workers intent on breaking down “information silos” across the federal government were trying to “unify systems into one central hub aims to advance multiple Trump administration priorities, including finding and deporting undocumented immigrants.” The documents note that the portal will also be accessible to so-called “non-Title 50” agencies outside of the national defense and intelligence apparatus.

Ayoub argued the intelligence community can’t provide access to its upcoming CAI portal without “raising the risk that agencies like DHS’s Homeland Security Investigations (HSI) would access the CAI database to identify and target noncitizens such as student protestors based on their search or browsing histories and location information.”

While the ODNI has acknowledged the importance of transparency, usernames for the portal will not include the name of the analyst’s agency, “thus obscuring any specific participation from individual participants,” according to the project documents.

“The irony is not lost on me that they are making efforts to protect individuals within the IC from being identified regarding their participation in this project but have no qualms about vacuuming up the personal data of Americans against their wishes and knowledge,” said Schroeder.

Sen. Ron Wyden, D-Ore., a longtime critic of the Fourth Amendment end run posed by CAI, expressed concern to The Intercept over how the portal will ultimately be used. “Policies are one thing, but I’m concerned about what the government is actually doing with data about Americans that it buys from data brokers,” he said in a statement. “All indications from news reports and Trump administration officials are that Americans should be extremely worried about how this administration may be using commercial data.”

The post U.S. Spy Agencies Are Getting a One-Stop Shop to Buy Your Most Sensitive Personal Data appeared first on The Intercept.

After President Donald Trump issued an executive order to rid the federal government of “diversity, equity, and inclusion” initiatives, the FCC in February launched a formal investigation into Verizon’s corporate diversity practices. The probe arrived as Verizon awaited approval for its planned $9.6 billion takeover of Frontier Communications. Carr has been open about leveraging the Frontier deal against Verizon. Asked in a CNBC interview whether he would block Verizon’s purchase if it didn’t shut down workplace diversity programs, he replied, “I’ve told everybody if they want to get a deal done before the FCC, they need to get rid of any invidious forms of discrimination.”

In a May 15 letter to Carr, obtained by The Intercept, Verizon chief legal officer Vandana Venkatesh writes that the company will comply with the government’s wishes. “Delivering for customers requires attracting the best talent from across the country,” Venkatesh says. “We are committed to creating a culture that leverages and values each person’s unique strengths and talents. These values have been fundamental to our Verizon culture since our founding 25 years ago. However, we recognize that the regulatory and policy landscape surrounding diversity, equity, and inclusion (“DEI”) has changed.” Following an internal evaluation, “Verizon recognizes that some DEl policies and practices could be associated with discrimination. For that reason, Verizon reaffirms its commitment to equal employment opportunity and nondiscrimination and is modifying its practices and ending its DEI-related policies.”

The changes include reassigning diversity-focused roles in Verizon’s human resources department; removing mentions of the words “diversity, equity, and inclusion” from employee training; ceasing to consider diversity when doing business with outside vendors in favor of contracting with “small businesses, including veteran-owned businesses”; and eliminating diversity-based hiring goals. The letter also says the company will cease certain “corporate sponsorships” in response to FCC demands — potentially a reference to the company’s past sponsorship of Pride events — and will cease to use the term “DEI” or the words “diversity, equity and inclusion” in any public messaging. Any prior reference to these words or concepts will be deleted, according to the letter.

The letter marks an about-face for Verizon and Venkatesh personally. In a 2023 LinkedIn post announcing a DEI-related company award, she wrote, “I’m so proud of the work the team has done to launch the annual IDEAward, recognizing law firms that share our commitment to promoting diversity, inclusion, equity and belonging in the legal industry. Congratulations to the winners and honorees of Verizon’s first annual IDEAward, and to the V-teamers who, like David, have dedicated their time and effort to making our company – and the legal profession – a more inclusive one.”

Verizon is just one in a string of corporate reversals on workplace diversity since Trump’s election. Amazon, Boeing, Disney, and many others have similarly shuttered diversity-related programs following years of right-wing pressure.

Verizon declined to comment.

The post Verizon Capitulates to Trump Administration, Cutting All DEI Programs appeared first on The Intercept.

The report makes explicit the extent to which the tech giant understood the risk of providing state-of-the-art cloud and machine learning tools to a nation long accused of systemic human rights violations and wartime atrocities. Not only would Google be unable to fully monitor or prevent Israel from using its software to harm Palestinians, but the report also notes that the contract could obligate Google to stonewall criminal investigations by other nations into Israel’s use of its technology. And it would require close collaboration with the Israeli security establishment — including joint drills and intelligence sharing — that was unprecedented in Google’s deals with other nations.

A third-party consultant Google hired to vet the deal recommended that the company withhold machine learning and artificial intelligence tools from Israel because of these risk factors.

Three international law experts who spoke with The Intercept said that Google’s awareness of the risks and foreknowledge that it could not conduct standard due diligence may pose legal liability for the company. The rarely discussed question of legal culpability has grown in significance as Israel enters the third year of what has widely been acknowledged as a genocide in Gaza — with shareholders pressing the company to conduct due diligence on whether its technology contributes to human rights abuses.

“They’re aware of the risk that their products might be used for rights violations,” said León Castellanos-Jankiewicz, a lawyer with the Asser Institute for International and European Law in The Hague, who reviewed portions of the report. “At the same time, they will have limited ability to identify and ultimately mitigate these risks.”

Google declined to answer any of a list of detailed questions sent by The Intercept about the company’s visibility into Israel’s use of its services or what control it has over Project Nimbus.

Company spokesperson Denise Duffy-Parkes instead responded with a verbatim copy of a statement that Google provided for a different article last year. “We’ve been very clear about the Nimbus contract, what it’s directed to, and the Terms of Service and Acceptable Use Policy that govern it. Nothing has changed.”

Portions of the internal document were first reported by the New York Times, but Google’s acknowledged inability to oversee Israel’s usage of its tools has not previously been disclosed.

In January 2021, just three months before Google won the Nimbus contract alongside Amazon, the company’s cloud computing executives faced a dilemma.

The Project Nimbus contract — then code-named “Selenite” at Google — was a clear moneymaker. According to the report, which provides an assessment of the risks and rewards of this venture, Google estimated a bespoke cloud data center for Israel, subject to Israeli sovereignty and law, could reap $3.3 billion between 2023 and 2027, not only by selling to Israel’s military but also its financial sector and corporations like pharmaceutical giant Teva.

But given decades of transgressions against international law by Israeli military and intelligence forces it was now supplying, the company acknowledged that the deal was not without peril. “Google Cloud Services could be used for, or linked to, the facilitation of human rights violations, including Israeli activity in the West Bank,” resulting in “reputation harm,” the company warned.

In the report, Google acknowledged the urgency of mitigating these risks, both to the human rights of Palestinians and Google’s public image, through due diligence and enforcement of the company’s terms of service, which forbid certain acts of destruction and criminality.

But the report makes clear a profound obstacle to any attempt at oversight: The Project Nimbus contract is written in such a way that Google would be largely kept in the dark about what exactly its customer was up to, and should any abuses ever come to light, obstructed from doing anything about them.

The document lays out the limitations in stark terms.

Google would only be given “very limited visibility” into how its software would be used. The company was “not permitted to restrict the types of services and information that the Government (including the Ministry of Defense and Israeli Security Agency) chooses to migrate” to the cloud.

Attempts to prevent Israeli military or spy agencies from using Google Cloud in ways damaging to Google “may be constrained by the terms of the tender, as Customers are entitled to use services for any reason except violation of applicable law to the Customer,” the document says. A later section of the report notes Project Nimbus would be under the exclusive legal jurisdiction of Israel, which, like the United States, is not a party to the Rome Statute and does not recognize the International Criminal Court.

Should Project Nimbus fall under legal scrutiny outside of Israel, Google is required to notify the Israeli government as early as possible, and must “Reject, Appeal, and Resist Foreign Government Access Requests.”

Google noted this could put the company at odds with foreign governments should they attempt to investigate Project Nimbus. The contract requires Google to “implement bespoke and strict processes to protect sensitive Government data,” according to a subsequent internal report, also viewed by The Intercept that was drafted after the company won its bid. This obligation would stand even if it means violating the law: “Google must not respond to law enforcement disclosure requests without consultation and in some cases approval from the Israeli authorities, which could cause us to breach international legal orders / law.”

The second report notes another onerous condition of the Nimbus deal: Israel “can extend the contract up to 23 years, with limited ability for Google to walk away.”

The initial report notes that Google Cloud chief Thomas Kurian would personally approve the contract with full understanding and acceptance of these risks before the company submitted its contract proposal. Google did not make Kurian available for comment.

Business for Social Responsibility, a human rights consultancy tapped by Google to vet the deal, recommended the company withhold machine learning and AI technologies specifically from the Israeli military in order to reduce potential harms, the document notes. It’s unclear how the company could have heeded this advice considering the limitations in the contract. The Intercept in 2022 reported that Google Cloud’s full suite of AI tools was made available to Israeli state customers, including the Ministry of Defense.

BSR did not respond to a request for comment.

The first internal Google report makes clear that the company worried how Israel might use its technology. “If Google Cloud moves forward with the tender, we recommend the business secure additional assurances to avoid Google Cloud services being used for, or linked to, the facilitation of human rights violations.”

It’s unclear if such assurances were ever offered.

Related

U.S. Companies Honed Their Surveillance Tech in Israel. Now It’s Coming Home.

Google has long defended Project Nimbus by stating that the contract “is not directed at highly sensitive, classified or military workloads relevant to weapons or intelligence services.” The internal materials note that Project Nimbus will entail nonclassified workloads from both the Ministry of Defense and Shin Bet, the country’s rough equivalent of the FBI. Classified workloads, one report states, will be handled by a second, separate contract code-named “Natrolite.” Google did not respond when asked about its involvement in the classified Natrolite project.

Both documents spell out that Project Nimbus entails a deep collaboration between Google and the Israeli security state through the creation of a Classified Team within Google. This team is made up of Israeli nationals within the company with security clearances, designed to “receive information by [Israel] that cannot be shared with [Google].” Google’s Classified Team “will participate in specialized training with government security agencies,” the first report states, as well as “joint drills and scenarios tailored to specific threats.”

The level of cooperation between Google and the Israeli security state appears to have been unprecedented at the time of the report. “The sensitivity of the information shared, and general working model for providing it to a government agency, is not currently provided to any country by GCP,” the first document says.

Whether Google could ever pull the plug on Nimbus for violating the company rules or the law is unclear. The company has claimed to The Intercept and other outlets that Project Nimbus is subject to its standard terms of use, like any other Google Cloud customer. But Israeli government documents contradict this, showing the use of Project Nimbus services is constrained not by Google’s normal terms, but a secret amended policy.

A spokesperson for the Israeli Ministry of Finance confirmed to The Intercept that the amended Project Nimbus terms of use are confidential. Shortly after Google won the Nimbus contract, an attorney from the Israeli Ministry of Finance, which oversaw the deal, was asked by reporters if the company could ever terminate service to the government. “According to the tender requirements, the answer is no,” he replied.

In its statement, Google points to a separate set of rules, its Acceptable Use Policy, that it says Israel must abide by. These rules prohibit actions that “violate or encourage the violation of the legal rights of others.” But the follow-up internal report suggests this Acceptable Use Policy is geared toward blocking illegal content like sexual imagery or computer viruses, not thwarting human rights abuses. Before the government agreed to abide by the AUP, Google wrote there was a “relatively low risk” of Israel violating the policy “as the Israel government should not be posting harmful content itself.” The second internal report also says that “if there is a conflict between Google’s terms” and the government’s requirements, “which are extensive and often ambiguous,” then “they will be interpreted in the way which is the most advantageous to the customer.”

International law is murky when it comes to the liability Google could face for supplying software to a government widely accused of committing a genocide and responsible for the occupation of the West Bank that is near-universally considered illegal.

Related

Google Won’t Say Anything About Israel Using Its Photo Software to Create Gaza “Hit List”

Legal culpability grows more ambiguous the farther you get from the actual act of killing. Google doesn’t furnish weapons to the military, but it provides computing services that allow the military to function — its ultimate function being, of course, the lethal use of those weapons. Under international law, only countries, not corporations, have binding human rights obligations. But if Project Nimbus were to be tied directly to the facilitation of a war crime or other crime against humanity, Google executives could hypothetically face criminal liability under customary international law or through a body like the ICC, which has jurisdiction in both the West Bank and Gaza.

Civil lawsuits are another option: Castellanos-Jankiewicz imagined a scenario in which a hypothetical plaintiff with access to the U.S. court system could sue Google over Project Nimbus for monetary damages, for example.

Along with its work for the Israeli military, Google through Project Nimbus sells cloud services to Israel Aerospace Industries, the state-owned weapons maker whose munitions have helped devastate Gaza. Another confirmed Project Nimbus customer is the Israel Land Authority, a state agency that among other responsibilities distributes parcels of land in the illegally annexed and occupied West Bank.

An October 2024 judicial opinion issued by the International Court of Justice, which arbitrates disputes between United Nations member states, urged countries to “take all reasonable measures” to prevent corporations from doing anything that might aid the illegal occupation of the West Bank. While nonbinding, “The advisory opinions of the International Court of Justice are generally perceived to be quite authoritative,” Ioannis Kalpouzos, a visiting professor at Harvard Law and expert on human rights law and laws of war, told The Intercept.

Establishing Google’s legal culpability in connection with the occupation of the West Bank or ongoing killing in Gaza entails a complex legal calculus, experts explained, hinging on the extent of its knowledge about how its products would be used (or abused), the foreseeability of crimes facilitated by those products, and how directly they contributed to the perpetration of the crimes. “Both the very existence of the document and the language used suggest at least the awareness of the likelihood of violations,” Kalpouzos said.

While there have been a few instances of corporate executives facing local criminal charges in connections with human rights atrocities, liability stemming from a civil lawsuit is more likely, said Castellanos-Jankiewicz. A hypothetical plaintiff might have a case if they could demonstrate that “Google knew or should have known that there was a risk that this software was going to be used or is being used,” he explained, “in the commission of serious human rights violations, war crimes, crimes against humanity, or genocide.”

Getting their date in court before an American judge, however, would be another matter. The 1789 Alien Tort Statute allows federal courts in the United States to take on lawsuits by foreign nationals regarding alleged violations of international law but has been narrowed considerably over the years, and whether U.S. corporations could even be sued under the statute in the first place remains undecided.

History has seen scant few examples of corporate accountability in connection with crimes against humanity. In 2004, IBM Germany donated $4 million to a Holocaust reparations fund in connection with its wartime role supplying computing services to the Third Reich. In the early 2000s, plaintiffs in the U.S. sued dozens of multinational corporations for their work with apartheid South Africa, including the sale of ”essential tools and services,” Castellanos-Jankiewicz told The Intercept, though these suits were thrown out following a 2016 Supreme Court decision. Most recently Lafarge, a French cement company, pleaded guilty in both the U.S. and France following criminal investigations into its business in ISIS-controlled Syria.

There is essentially no legal precedent as to whether the provision of software to a military committing atrocities makes the software company complicit in those acts. For any court potentially reviewing this, an important legal standard, Castellanos-Jankiewicz said, is whether “Google knew or should have known that its equipment that its software was being either used to commit the atrocities or enabling the commission of the atrocities.”

The Nimbus deal was inked before Hamas attacked Israel on October 7, 2023, igniting a war that has killed tens of thousands of civilians and reduced Gaza to rubble. But that doesn’t mean the company wouldn’t face scrutiny for continuing to provide service. “If the risk of misuse of a technology grows over time, the company needs to react accordingly,” said Andreas Schüller, co-director of the international crimes and accountability program at the European Center for Constitutional and Human Rights. “Ignorance and an omission of any form of reaction to an increasing risk in connection with the use of the product leads to a higher liability risk for the company.”

Though corporations are generally exempt from human rights obligations under international frameworks, Google says it adheres to the United Nations Guiding Principles on Business and Human Rights. The document, while voluntary and not legally binding, lays out an array of practices multinational corporations should follow to avoid culpability in human rights violations.

Among these corporate responsibilities is “assessing actual and potential human rights impacts, integrating and acting upon the findings, tracking responses, and communicating how impacts are addressed.”

The board of directors at Alphabet, Google’s parent entity, recently recommended voting against a shareholder proposal to conduct an independent third-party audit of the processes the company uses “to determine whether customers’ use of products and services for surveillance, censorship, and/or military purposes contributes to human rights harms in conflict-affected and high-risk areas.” The proposal cites, among other risk areas, the Project Nimbus contract. In rejecting the proposal, the board touted its existing human rights oversight processes, and cites the U.N. Guiding Principles and Google’s “AI Principles” as reason no further oversight is necessary. In February, Google amended this latter document to remove prohibitions against weapons and surveillance.

“The UN guiding principles, plain and simple, require companies to conduct due diligence,” said Castellanos-Jankiewicz. “Google acknowledging that they will not be able to conduct these screenings periodically flies against the whole idea of due diligence. It sounds like Google is giving the Israeli military a blank check to basically use their technology for whatever they want.”

The post Google Worried It Couldn’t Control How Israel Uses Project Nimbus, Files Reveal appeared first on The Intercept.