Hegseth gives Anthopic ultimatum on guardrails

What happened

Defense Secretary Pete Hegseth Tuesday gave Anthropic until Friday to allow the Pentagon unrestricted access to its Claude artificial intelligence tool or face serious consequences.

In a “tense meeting,” Axios said, Hegseth told Anthropic CEO Dario Amodei that if his firm did not drop its safety guardrails, the Pentagon would cancel its $200 million contract and “declare Anthropic a ‘supply chain risk,’ or invoke the Defense Production Act to force the company to tailor its model to the military’s needs.”

Who said what

Anthropic has “aggressively positioned itself to be a key player in national security” and was the first AI firm to “integrate its technology into the Pentagon’s classified networks,” The Washington Post said. But tensions have mounted over Amodei’s “ethical concerns about unchecked government use of AI,” The Associated Press said, especially for “fully autonomous armed drones” and “AI-assisted mass surveillance that could track dissent.” Some Trump administration officials have lambasted Anthropic’s “hard-line on domestic surveillance and AI weapons” as “Woke AI,” NPR said.

Tuesday’s meeting “ended in a stalemate,” as Amodei “reiterated the company’s red lines and said they wouldn’t interfere with the Pentagon’s operations,” The Wall Street Journal said. Either of Hegseth’s threatened sanctions “would be nearly unprecedented,” and a supply-chain risk designation would affect “a swath of companies, including many in the tech sector.” Defense Production Act experts “questioned whether it could be used to force Anthropic to drop the limitations it seeks to maintain,” the Post said.

What next?

The Pentagon’s threats are “extreme,” legally questionable and “strategically counterproductive,” Lawfare said. But the “deeper problem” is that the “terms governing how the military uses the most transformative technology of the century are being set through bilateral haggling between a defense secretary and a startup CEO,” not by Congress.