OpenAI Inks Pentagon Deal After Anthropic Cut
OpenAI reached a new agreement with the U.S. Department of Defense, signaling a major push into the government sector. The deal came just hours after a similar contract with rival Anthropic was reportedly cut over policy disputes. CEO Sam Altman confirmed ongoing negotiations, highlighting the unique compliance and security demands of defense contracts.
The deal follows OpenAI's quiet removal of language from its usage policy in January 2026 that had previously banned "military and warfare" applications. The company maintained a prohibition on using its AI to develop weapons or cause harm, but the change cleared a path for broader national security collaborations, such as cybersecurity tools being developed with DARPA. Anthropic's contract dispute centered on its refusal to remove safeguards preventing its AI, Claude, from being used for mass domestic surveillance or in fully autonomous weapons systems without human oversight. The Pentagon demanded the model be available for "all lawful purposes" and, after negotiations failed, designated Anthropic a "supply-chain risk to national security," a move typically reserved for foreign adversaries. The agreement allows for the deployment of OpenAI's models on the Pentagon's classified networks. However, reports indicate OpenAI currently lacks the necessary infrastructure and security certifications for classified work, a process that can take months or even years to complete. In contrast, Anthropic's Claude was already operational on the military's secure cloud systems through a partnership with Palantir. OpenAI CEO Sam Altman stated the Pentagon deal includes prohibitions on domestic mass surveillance and requires human responsibility for the use of force, principles he said are reflected in law and policy. These were the same "red lines" that Anthropic publicly championed, leading to its contract termination. The Pentagon's aggressive stance is part of a broader push to accelerate AI adoption and maintain a technological edge, outlined in its "AI-First" strategy. This strategy emphasizes rapid experimentation with commercial AI and requires that the latest models be deployable within 30 days of public release. The Chief Digital and Artificial Intelligence Office (CDAO) is overseeing these efforts. This deal is part of a larger competitive landscape where major AI firms are vying for lucrative defense contracts. Last summer, the Pentagon awarded contracts worth up to $200 million each to OpenAI, Anthropic, Google, and xAI for "frontier AI projects". The intense competition for these high-value government deals is reshaping how AI companies approach ethics and policy.
