Altman Admits Pentagon Deal Looked 'Sloppy'
OpenAI CEO Sam Altman has acknowledged that the company's exclusive Pentagon deal, struck after the government banned competitor Anthropic, "looked opportunistic and sloppy." The comment fuels criticism over the deal's haste and lack of transparency, highlighting the reputational risks for AI firms navigating high-stakes government contracts in the national security space.
The controversial deal followed the Trump administration's abrupt blacklisting of competitor Anthropic on February 28, 2026. The Pentagon declared Anthropic a "supply-chain risk" and ordered all federal agencies to cease using its technology after the company refused to drop contractual "red lines." Those red lines included prohibitions on using its AI for mass domestic surveillance and for fully autonomous weapons without human control. Anthropic's CEO, Dario Amodei, stated the government's action felt "punitive and inappropriate," especially since the company's technology had been used in classified Pentagon work for the past year. The ban, which canceled over $200 million in federal contracts for Anthropic, was seen by many as retaliation for the company upholding its ethical guidelines. In the wake of Anthropic's blacklisting, OpenAI announced its own agreement to provide AI models for the Pentagon's classified networks. This move triggered significant backlash, with a surge of users reportedly canceling their ChatGPT subscriptions and switching to Anthropic's Claude. Facing public pressure and criticism from within, OpenAI amended its Pentagon contract. The revised agreement explicitly bans the use of its AI for intentional domestic surveillance of U.S. citizens and restricts intelligence agencies like the NSA from using the technology without a separate contract modification. Initially, OpenAI's usage policy, updated in January 2024, had removed explicit bans on "military and warfare," which contrasted with Anthropic's steadfast position. Despite the controversy, OpenAI stated it negotiated for similar contractual terms to be offered to other AI labs in an attempt to "de-escalate" the situation between the U.S. government and AI companies. The episode has highlighted the intense debate over the ethical responsibilities of AI developers when contracting with military and intelligence agencies. Even with the amended contract, some critics argue that the "lawful purposes" language in OpenAI's deal remains ambiguous and contains loopholes. Internal dissent has also surfaced, with some OpenAI employees expressing concern that the company compromised its safety principles. A former policy research head at OpenAI suggested the company "caved + framed it as not caving, and screwed Anthropic while framing it as helping them."
