OpenAI Renegotiates 'Sloppy' Pentagon Deal
OpenAI CEO Sam Altman says the company is renegotiating its defense contract with the Pentagon, calling the original deal "opportunistic and sloppy." The move aims to set clearer ethical boundaries amid rising enterprise security concerns, which may be reflected in a reported 295% spike in ChatGPT uninstalls following a DoD deal.
The swift renegotiation follows a contentious backdrop where rival AI firm Anthropic was dropped by the Pentagon. Anthropic had refused to remove safeguards preventing its AI, Claude, from being used for mass domestic surveillance or fully autonomous weapons, leading the Trump administration to designate it a "supply chain risk." The original OpenAI agreement, announced almost immediately after Anthropic's ousting, permitted use for all "lawful purposes," raising alarms that it could allow for sweeping domestic spying under existing legal frameworks. Critics argued these broad terms mirrored the legal justifications used for controversial programs like the NSA's PRISM data collection. In response to the backlash, the amended contract now explicitly forbids using OpenAI's tools for "domestic surveillance of US persons and nationals," including through commercially purchased data. Furthermore, the deal bars use by intelligence agencies like the NSA, which would now require a separate contract modification to gain access. This deal is part of a wider Pentagon strategy to integrate advanced AI, with initial contracts of up to $200 million each also awarded to Google and xAI. OpenAI's initial contract, valued at up to $200 million, is for developing "prototype frontier AI capabilities" for both warfighting and enterprise functions, with a projected completion date of July 2026. OpenAI's willingness to engage with the military follows a notable policy shift in January 2024. The company quietly removed a blanket ban on "military and warfare" from its usage policy, replacing it with a more specific prohibition against using its tech to cause harm or develop weapons. The controversy has had immediate market repercussions. Following the initial deal's announcement, Anthropic's Claude chatbot surged to the top of Apple's App Store charts, surpassing ChatGPT. Sam Altman has since publicly stated that he urged the Pentagon to offer the same revised, more restrictive terms to Anthropic and other AI labs.
