Anthropic faces a Friday evening deadline to grant the U.S. military unrestricted access to its AI model, as reported by Axios. Defense Secretary Pete Hegseth informed Anthropic CEO Dario Amodei that the Pentagon may either label Anthropic a “supply chain risk” or use the Defense Production Act (DPA) to adapt the model for military use. The DPA allows the president to compel companies to prioritize production for national defense, as seen during the COVID-19 pandemic with companies like General Motors and 3M. Anthropic opposes its technology’s use in mass surveillance or autonomous weapons and won’t compromise on these issues.
Pentagon officials assert that military technology use should follow U.S. law and constitutional restrictions, not private contractor policies. Employing the DPA in an AI dispute could broaden its modern application and signal increased executive branch instability, says Dean Ball of the Foundation for American Innovation. Ball warns this could be perceived as the government pressuring dissenting businesses politically.
Some administration members, including AI czar David Sacks, criticize Anthropic’s safety policies as “woke.” Ball suggests this dispute could deter businesses, fearing instability in the U.S. legal system.
Anthropic appears unlikely to concede, according to Reuters. Currently, it is the only frontier AI lab with classified DOD access, while the Pentagon has struck a deal for xAI’s Grok for classified systems. Ball explains the lack of options for the DOD may justify its aggressive approach; the agency lacks backup and risks not fulfilling a National Security Memorandum guideline to avoid dependency on a single AI system. TechCrunch has contacted Anthropic and the DOD for comments.