OpenAI Deploys in Pentagon, Stirs Dissent
OpenAI CEO Sam Altman confirmed the company's technology is now deployed inside Pentagon systems, pledging deference to government oversight. The move is reportedly causing significant internal backlash, with some staff fuming over the company's deepening military ties and the ethical implications of their work.
The deal followed the U.S. government designating rival AI firm Anthropic a "supply chain risk" and ordering federal agencies to stop using its technology. Anthropic's blacklisting came after it refused to remove contractual red lines prohibiting the use of its models for fully autonomous weapons and mass domestic surveillance. OpenAI's agreement accepts a "lawful purposes" framework that Anthropic rejected, arguing the legal definitions have not caught up to AI's capabilities for surveillance at scale. This move represents a significant policy evolution for OpenAI, which in January 2024 removed language from its usage policy that explicitly banned "military and warfare" applications. The Pentagon's Chief Digital and AI Office (CDAO) awarded OpenAI a $200 million prototype agreement to explore "frontier AI capabilities" for both warfighting and enterprise domains, with a completion date of July 2026. This is part of a broader "AI-first" strategy to accelerate experimentation and deployment of advanced AI tools across the Department of Defense. Internal dissent at OpenAI became public, with some researchers posting on X that the deal wasn't "worth it" and calling for independent legal counsel to review the terms. Nearly 900 employees from OpenAI and Google also signed an open letter urging their companies to reject demands related to surveillance and autonomous weapons. The public backlash was immediate, with uninstalls of the ChatGPT mobile app reportedly surging 295% in one day, while downloads for Anthropic's Claude chatbot climbed to the top of the App Store rankings. In response to the criticism, CEO Sam Altman admitted the rollout "looked opportunistic and sloppy" and announced revisions to the contract. The amended terms now explicitly prohibit the intentional use of its AI for domestic surveillance of U.S. persons and restrict intelligence agencies like the NSA from using the systems without a separate contract modification.
