- Anthropic rejects Pentagon terms over AI safeguards dispute.
- Company refuses use in autonomous weapons, mass surveillance.
- Contract worth up to $200 million at risk.
- Rivals accept "all lawful purposes" Pentagon agreements.
Anthropic chief executive Dario Amodei said on Thursday the company cannot accept the Pentagon's terms in a deepening dispute over AI safeguards, rejecting what defence officials called their last and final offer and defying a Friday 5:01 p.m. ET deadline for compliance.
At stake is a contract worth up to $200 million and Anthropic's continued access to classified US military networks, where Claude has been the only large language model deployed to date.
Two Redlines: Autonomous Weapons and Mass Surveillance of Americans
The dispute centres on two limits Anthropic refuses to drop: a prohibition on Claude powering fully autonomous weapons systems that select and kill targets without human authorization and a bar on mass domestic surveillance of Americans.
The Pentagon requires all AI contractors to permit models for all lawful purposes in classified settings. "The contract language we received overnight from the Department of War made virtually no progress on preventing Claude's use for mass surveillance of Americans or in fully autonomous weapons," Anthropic said.
Also Read: NVIDIA CEO Announces Agentic AI Inflection Point; Tech Race Remodeling Nations
"New language framed as compromise was paired with legalese that would allow those safeguards to be disregarded at will."
Hegseth Threatens Blacklist; Pentagon Official Calls Amodei a 'Liar'
The standoff escalated Tuesday when Defence Secretary Pete Hegseth met Amodei at the Pentagon, threatening to cancel the contract, designate Anthropic a supply chain risk a label reserved for companies linked to foreign adversaries and invoke the Defence Production Act to compel compliance.
Pentagon spokesman Sean Parnell wrote on X: "Here's what we're asking: Allow the Pentagon to use Anthropic's model for all lawful purposes." Undersecretary Emil Michael separately posted: "It's a shame that @DarioAmodei is a liar and has a God-complex."
Amodei Holds Firm; Rivals Sign 'All Lawful Purposes' Deals With Pentagon
In a blog post, Amodei wrote: "I believe deeply in the existential importance of using AI to defend the United States and other democracies, and to defeat our autocratic adversaries." He added: "These threats do not change our position: we cannot in good conscience accede to their request." Should the Pentagon proceed, Amodei said the company "will work to enable a smooth transition to another provider."
An Anthropic spokesperson confirmed the company remains ready to continue negotiations. Rivals OpenAI, Google and xAI have each accepted all lawful purposes contracts; xAI extended that agreement to classified systems this week.
Bloomberg reported that in December, a senior Pentagon official posed a nuclear missile hypothetical to Amodei asking whether Claude's safeguards would block a response with 90 seconds to spare illustrating how wide the ideological gap between both sides remains.
