OpenAI's Pentagon Deal Hits Turbulence

The controversial deal is valued at up to $200 million and is intended to "develop prototype frontier AI" for both administrative and warfighting applications within the Defense Department. The project, managed by the Pentagon's Chief Digital & AI Officer (CDAO), has a projected completion date of July 2026. The timing of the announcement drew sharp criticism, coming just hours after the Trump administration ordered federal agencies to cease using AI from rival company Anthropic. Anthropic had reportedly refused to relax its safety standards regarding autonomous weapons and mass surveillance, leading to its designation as a "supply chain risk." In response to the backlash, OpenAI amended the agreement to include more explicit "red lines." The updated contract language specifically prohibits the intentional use of its AI for the domestic surveillance of U.S. persons and clarifies that intelligence agencies like the NSA cannot use the services without a new agreement. This partnership marks a significant evolution in OpenAI's policy, which previously had a broader ban on "military and warfare" applications. In January 2024, the company updated its usage policy to forbid using its services to "harm" people, including developing weapons, but removed the comprehensive prohibition on military collaborations. Sam Altman defended the deal by stating OpenAI was trying to "de-escalate things and avoid a much worse outcome," but acknowledged the poor timing. He also stated that he had communicated to the Pentagon that Anthropic should not be designated as a supply chain risk and should be offered the same terms. The agreement stipulates that OpenAI will retain control over its safety features and that cleared company personnel will be involved in the deployment on classified networks. The deal is part of a broader push by the CDAO, established in 2022, to accelerate the adoption of advanced AI across the armed services.