OpenAI Inks Pentagon Deal
Days after the White House banned rival Anthropic AI over security risks, OpenAI has secured a new contract with the Pentagon. CEO Sam Altman said the deal includes new “technical safeguards,” positioning OpenAI as the go-to AI provider for US defense. The move is a major coup for OpenAI and a stark warning to other AI firms that national security concerns can instantly reshape the competitive landscape.
The deal restricts OpenAI's technology from being used for mass domestic surveillance, to direct autonomous weapons, or for high-stakes automated decisions. To enforce this, the models will be deployed only on the cloud, not on "edge" devices like drones, and OpenAI will retain full control over its safety systems. The contract's language specifically references current laws and policies, meaning future changes to those laws won't override the agreed-upon protections. This agreement follows the White House's dramatic ban of rival AI company Anthropic. The administration designated Anthropic a "supply-chain risk" after the company refused to remove safeguards that prevented its AI, Claude, from being used for the very same purposes OpenAI's deal now prohibits. The Pentagon had given Anthropic a deadline to allow for "all lawful purposes" use, which the company stated it could not accept in good conscience. President Trump ordered all federal agencies to cease using Anthropic's technology, with the Pentagon given a six-month phase-out period. In a social media post, the former president accused the company of trying to "strong-arm" the Department of War and putting national security in jeopardy. Defense Secretary Pete Hegseth stated that no contractor doing business with the military may conduct any commercial activity with Anthropic. The Pentagon has been pursuing an "AI-first" warfighting force and has awarded contracts worth up to $200 million each to several major AI labs, including Google and xAI, in the past year. This is part of a broader "AI Acceleration Strategy" which includes seven "pace-setting projects" to fast-track capabilities in areas like AI-enabled combat concepts and intelligence analysis. The controversy highlights a fundamental tension between AI companies' safety concerns and the military's desire for unrestricted access to cutting-edge technology. OpenAI's deal suggests a possible compromise, with CEO Sam Altman stating the company asked the Pentagon to offer the same terms to all AI companies to "de-escalate" the conflict. This effort to integrate advanced AI into military operations builds on initiatives like Project Maven, started in 2017 to use machine learning to analyze drone surveillance footage. That project, now a formal Program of Record, has involved contractors like Palantir and aims to create a comprehensive multi-domain targeting system. OpenAI has publicly shared the specific contract language detailing its "red lines" and safeguards. The company also stated it would have cleared OpenAI engineers working with the government to ensure compliance and could terminate the contract if the terms were violated. Despite the deal, OpenAI has also expressed solidarity with its rival, stating publicly that Anthropic should not be labeled a "supply-chain risk." Around 70 OpenAI employees signed an open letter titled "We Will Not Be Divided," expressing support for Anthropic's stance.
