Anthropic CEO rejects Pentagon ultimatum over AI weapons safeguards

Anthropic CEO Dario Amodei said Thursday that "The Defense Department's threats do not change our position: we cannot in good conscience accede to their request" -- publicly rejecting the Pentagon's final offer to comply with military demands by Friday, February 28 at 5:01 PM. The Pentagon is demanding that Anthropic remove contractual safeguards barring its Claude AI from being used for mass surveillance of Americans and for autonomous weapons targeting decisions without human oversight.

The standoff exposes a direct collision between AI safety commitments and U.S. defense priorities. Defense Secretary Pete Hegseth issued the ultimatum Tuesday after meeting with Amodei, warning the company to "get with military program by Friday or lose $200 million." Amodei countered that "frontier AI systems are simply not reliable enough to power fully autonomous weapons" and that autonomous systems "cannot be relied upon to exercise the critical judgment that our highly trained, professional troops exhibit every day." If Anthropic misses the Friday deadline, it faces contract termination, a "supply chain risk" designation requiring all Pentagon vendors to certify non-use of Anthropic technology, and potential invocation of the Defense Production Act to compel access to Claude. Rival xAI has already complied with equivalent Pentagon terms for its Grok model, and Google is reported to be close to a similar classified-use deal.