Defense Secretary Pete Hegseth delivered an ultimatum to Anthropic, the artificial intelligence firm, on Tuesday, demanding full access to its Claude AI model by the conclude of this week, or face potential penalties, including the cancellation of a $200 million contract and designation as a “supply chain risk.” The demand, reported by CBS News and Axios, stems from disagreements over the military’s intended utilize of the AI and Anthropic’s attempts to impose safeguards against potential misuse.
The Pentagon seeks unfettered access to Claude’s capabilities for military operations, while Anthropic has resisted allowing its use for mass surveillance or in autonomous weapons systems, according to sources familiar with the negotiations. Dario Amodei, Anthropic’s CEO, was presented with the department’s terms during a meeting with Hegseth, and given until Friday to respond, Axios reported.
Anthropic, positioning itself as a leader in AI safety, has repeatedly requested that the Defense Department agree to restrictions preventing Claude from being used to surveil Americans, a practice officials have stated is already illegal. Amodei also expressed concerns about the AI being used for final targeting decisions in military operations without human oversight, citing the potential for “hallucinations” and errors that could lead to unintended escalation or mission failure, CBS News reported. A senior Pentagon official dismissed these concerns, stating that the requests are for lawful activities only.
The dispute highlights a growing tension between the U.S. Military’s desire to leverage the power of AI and the ethical considerations raised by AI developers. The Department of Defense has already integrated Claude into its operations, but views Anthropic’s restrictions as roadblocks. The Pentagon is also considering invoking the Defense Production Act to compel Anthropic’s compliance, CBS News reported.
The standoff could set a precedent for how the AI industry responds to government demands for military applications of their technology. Other AI companies, including xAI, owned by Elon Musk, are reportedly more willing to cooperate with the Pentagon in classified settings. According to the Pentagon official, xAI’s Grok is already being used in a classified capacity. Reuters reported that Anthropic has no intention of easing its usage restrictions for military purposes.
