President Trump broke bread with Anthropic CEO Dario Amodei at a private White House dinner Sunday — in what appeared to be a remarkable thaw just two days after a federal appeals court upheld a Pentagon move to exclude the artificial intelligence firm as a supply-chain risk.

The dinner marked the first reported one-on-one meeting between the commander-in-chief and the techie after months of bruising clashes over how the military can use Anthropic’s Claude models, according to Axios and the Wall Street Journal.

Trump personally invited Amodei after the AI guru missed last week’s White House state dinner because of a scheduling conflict, Axios reported.

The sit-down appeared to mark a striking shift for Amodei, whose company has spent much of the year battling the Trump administration in court — and whose views on AI safety have drawn fire from the president and his allies. As recently as Sept. 14, Trump mockingly called Amodei a “perfect little angel” in slamming his calls for AI guardrails.

Trump confirmed the meeting before it took place and described Amodei as highly respected, according to the Journal.

The president has nevertheless been making clear that he remains wary of calls to pump the brakes on AI development as the US races China for technological supremacy.

“We have to be a little careful,” he said in a recent “All In” podcast interview cited by the Journal, while dismissing dire warnings that AI could take over the world as a “hoax.”

“That doesn’t mean we are going to stop an industry,” Trump added.

Amodei has staked out a starkly different position on the risks posed by increasingly powerful AI.

Anthropic has pushed for greater oversight of frontier models, with Amodei recently calling for slowing the pace of capability growth — even as he has argued that the US needs to maintain a technological advantage over China.

The two sides’ biggest rupture to date came over the Pentagon’s use of Claude.

Anthropic had already allowed the government to use its technology for sensitive work including intelligence analysis, offensive cyber operations and the design of weapons systems, according to records in its ongoing case over the supply-chain risk designation.

Amodei previously refused a Pentagon demand that Claude be made available for “all lawful uses,” citing two red lines: mass domestic surveillance and fully autonomous lethal weapons.

The Pentagon responded by moving to exclude Anthropic from its supply chain.

That fight has since produced two separate court rulings that cut in different directions.

A federal judge in California ruled in August that a War Department designation under one supply-chain statute was unlawful, finding that Anthropic did not meet that law’s definition of a supply-chain risk.

But on Friday, a divided US Court of Appeals for the DC Circuit upheld a separate Pentagon action under the Federal Acquisition Supply Chain Security Act.

The 2-1 majority concluded that Anthropic’s ability to build restrictions into Claude — along with documented instances in which the model refused government requests — could create the kind of operational risk covered by that law.

The Post has sought comment from the White House and Anthropic.