Dario Amodei has walked back his leaked internal memo criticizing the Trump administration, calling it a heat-of-the-moment post that he dashed off within hours of one of the worst news days in Anthropic’s short history—a Friday afternoon in which Trump threatened to pull the company from all federal systems, the Pentagon announced a supply chain risk designation, and OpenAI swooped in with its own Pentagon deal, all before the day was out.“It was a difficult day for the company, and I apologize for the tone of the post,” Amodei wrote in a statement published on Anthropic’s website Thursday. “It does not reflect my careful or considered views. It was also written six days ago, and is an out-of-date assessment of the current situation.”

Pentagon officially designates Anthropic a supply chain risk, the first time a US company has received such a label

The apology came the same day the Pentagon formally notified Anthropic that it and its products are “deemed a supply chain risk, effective immediately”—a designation typically reserved for companies tied to foreign adversaries. Defense Secretary Pete Hegseth had telegraphed the move last week, but both sides had continued talking behind the scenes in hopes of a deal. Those talks appear to have collapsed, at least for now.The memo that set everything off was first reported by The Information. In it, Amodei told staff that the Trump administration’s hostility toward Anthropic was because the company hadn’t donated to the president and hadn’t offered what he called “dictator-style praise”—a jab clearly aimed at OpenAI CEO Sam Altman, who donated $1 million to Trump’s inauguration fund and whose company quickly signed a deal with the Pentagon after Anthropic’s talks broke down.Amodei says he wrote the memo minutes after Trump and Hegseth announced measures that could gut Anthropic’s government businessIn his Anthropic blog post, Amodei explained that the memo was written immediately after Trump posted on Truth Social about removing Anthropic from federal systems, Hegseth announced the supply chain risk designation on X, and OpenAI’s Pentagon deal dropped—all in the same news cycle. In that context, he argued, matters.“Anthropic did not leak this post nor direct anyone else to do so,” he added.Despite the apology, Anthropic is not backing down legally. The company says it will challenge the Pentagon’s designation in court, arguing it isn’t legally sound under the relevant statute—10 USC 3252—which requires the Secretary of War to use the least restrictive means necessary. Amodei also reiterated that Anthropic’s two core objections remain unchanged: no fully autonomous weapons and no mass domestic surveillance.The Pentagon was still actively using Claude to support military operations, including in Iran, adding a surreal dimension to a standoff that shows no signs of a clean resolution.