OpenAI's Pentagon Deal Called 'Sloppy'
OpenAI CEO Sam Altman now admits the company's high-profile Pentagon deal looked "opportunistic and sloppy" in its rollout. The comment comes as reports reveal OpenAI's willingness to rapidly amend its policies to allow for classified and intelligence applications was key to securing the contract, reigniting debates over AI militarization.
The controversial deal followed the Pentagon's blacklisting of rival AI company Anthropic, which had refused a contract that didn't explicitly forbid the use of its technology for mass domestic surveillance or autonomous weapons. After Anthropic held its ground, the Trump administration designated it a "supply chain risk," a move typically reserved for foreign adversaries. OpenAI's agreement, valued at up to $200 million, is part of a new "OpenAI For Government" initiative. The project aims to develop "prototype frontier AI" for both administrative functions and warfighting operations, with an estimated completion date of July 2026. The initial backlash against OpenAI was swift, with a "delete ChatGPT" campaign trending online and competitor Claude reportedly surging to the top of app store charts. In response to the criticism, OpenAI amended the contract to explicitly ban its AI from being used for domestic surveillance of U.S. citizens and barred intelligence agencies like the NSA from using its services. OpenAI had removed a general ban on "military and warfare" applications from its usage policy in January 2024, well before the Pentagon deal was announced. This policy shift paved the way for the company to engage with military contracts. The incident sparked a broader debate within the tech community, with hundreds of employees from major companies like Google, Microsoft, and Amazon signing an open letter. They urged their employers to reject demands from the Pentagon that could lead to the use of AI for surveillance and autonomous killing. Numerous other tech firms have long-standing relationships with the defense sector, including Palantir, Lockheed Martin, and Anduril. These companies provide a range of AI-powered services from battlefield analytics and autonomous navigation for drones to predictive maintenance and cyber threat detection.
