Pete Hegseth wants a top Pentagon contractor to know he really, really means it when he demands they abandon their cautious approach to the Defense Department’s AI systems.
The defense secretary is set to host Anthropic CEO Dario Amodei on Tuesday morning for what Axios reports is “likely to be tense meeting” on the military’s use of the company’s Claude software.
“Anthropic knows this is not a get-to-know-you meeting,” as one defense official described it. “This is not a friendly meeting. This is a s--t-or-get-off-the-pot-meeting.”
Amodei, who’s often warned of AI’s potential for misuse, has consistently pushed to frame his firm as a safety-conscious leader in the sector.
He has so far resisted pressure from Hegseth to remove safeguards on the Pentagon’s Claude-enabled programs unless the department agrees to wall off mass surveillance of citizens and research into weapons capable of firing without a human operator.
Their feud is understood to have escalated amid reports that Claude was used by the Pentagon in the Trump administration’s lightning invasion of Venezuela earlier in January.
Critics decried that mission—which secured the capture of President Nicolas Maduro, who now faces narcoterrorism charges in New York federal court—as an all-out assault on the rules-based international order.
Anthropic told Axios that “we are having productive conversations, in good faith,” with the Pentagon.
Defense officials instead say “negotiations have shown no progress,” and are now “on the verge of breaking down.”
Earlier this month, Hegseth warned he would consider certifying Anthropic as a “supply chain risk” if it did not yield to his demands.
Officials said that at the Tuesday meeting, the secretary now plans on issuing Amodei with an “ultimatum.”
“The problem with Dario is, with him, it’s ideological,” one senior Defense Department source said. “We know who we’re dealing with.”
The Daily Beast has contacted the Defense Department and Anthropic for comment on this story.
