Warren Demands Full Disclosure of Pentagon AI Contracts

Senator Elizabeth Warren (D-Mass.) sent letters to the Pentagon and seven major AI companies on Monday demanding the full disclosure of military AI contracts, escalating her year-long investigation into the Defense Department’s opaque artificial intelligence arrangements. The letters, sent to Google, OpenAI, Microsoft, Amazon Web Services, Nvidia, SpaceX, and startup Reflection AI, seek transparency about agreements that allow these companies to deploy AI models on classified military networks handling secret and top-secret data.

Background: The Pentagon’s AI Push

The demand follows the Pentagon’s May 1 announcement that it had reached classified-network agreements with seven tech companies through its GenAI.mil portal, which already serves 1.3 million military users who have built “hundreds of thousands” of AI agents. According to NBC News, the contracts cover “warfighting, intelligence and enterprise operations” — including analyzing intelligence, identifying targets, organizing logistics, and making battlefield decisions.

The Associated Press reported that the Pentagon’s chief technology officer, Emil Michael, defended the multi-provider approach, stating that “when we learned that one partner didn’t really want to work with us in the way we wanted to work with them, we went out and made sure that we had multiple different providers.”

The Anthropic Controversy

At the center of the dispute is AI safety company Anthropic, which refused to accept the Pentagon’s “all lawful” use clause — the same language the other seven companies accepted. Anthropic sought contractual guardrails preventing mass surveillance of Americans, fully autonomous lethal weapons, and warrantless data broker purchases. In response, the Pentagon designated Anthropic a “supply chain risk,” effectively blacklisting the company from federal contracts. Anthropic CEO Dario Amodei called warrantless AI-powered mass surveillance “a crime against humanity.”

Surveillance Concerns and Loopholes

OpenAI, which reversed its 2023 ban on military use and signed a $200 million contract with the Pentagon, was the only company to publicly negotiate surveillance restrictions. The revised agreement includes language that the AI “shall not be intentionally used for domestic surveillance of U.S. persons and nationals.” However, as State of Surveillance detailed, significant loopholes remain: intelligence agencies can access the models with a separate contract modification, the word “intentionally” does not cover incidental collection under programs like Section 702 FISA, and the revised agreement had not been formally signed as of reporting.

Warren has previously expressed concern that “DoD is trying to strong-arm American companies into providing the Department with the tools to spy on American citizens and deploy fully autonomous weapons without adequate safeguards.”

Warren’s History on the Issue

The senator has been investigating Pentagon AI contracting for over a year. She questioned the Pentagon over a $200 million xAI/Grok contract in September 2025, raised concerns about classified access to Elon Musk’s Grok AI in March 2026, and opened an investigation into the Pentagon’s designation of Anthropic as a supply chain risk later that same month. She has also co-sponsored the bipartisan Protecting AI Competition Act with Sen. Eric Schmitt (R-Mo.).

Expert Analysis

Helen Toner of Georgetown University’s Center for Security and Emerging Technology noted that while AI systems can be helpful for summarizing information and identifying potential targets, questions about appropriate human involvement remain unresolved. “There’s a phenomenon called automation bias, where people can be prone to assume that machines work better than they actually do,” Toner told the AP.

The Broader Implications

The Pentagon’s approach — rewarding companies that accept “all lawful” use while punishing those that negotiate restrictions — sets a powerful precedent for the entire defense AI industry. The inclusion of Reflection AI, a startup with almost no public information about its capabilities or leadership, alongside established tech giants raises further questions about the vetting process for companies receiving classified network access.

Warren’s push also highlights a growing bipartisan dimension to military AI oversight. Her previous collaboration with Republican Sen. Eric Schmitt on the Protecting AI Competition Act suggests potential for cross-party support on transparency measures, even as the Trump administration’s Pentagon has moved aggressively to expand AI capabilities on classified networks.

What’s Next

The Pentagon now faces a choice between complying with Warren’s transparency demands and maintaining the classification of these agreements. As a member of the Senate Armed Services Committee, Warren has institutional leverage to compel disclosure, but the contracts govern AI systems on secret and top-secret networks, making public oversight extremely difficult. The response from the seven companies — and whether Congress will take legislative action on military AI oversight — will determine the next chapter in this unfolding debate over the balance between national security and public accountability.