IT·Science News

Trump bars agencies from using Anthropic’s AI, and OpenAI signs a Pentagon deal hours later

President Donald Trump on Friday 27 February ordered all US federal agencies to stop using Anthropic’s AI, and Defense Secretary Pete Hegseth designated the company a “supply chain risk”, the Associated Press reports via Federal News Network. The label is normally used for firms tied to US adversaries. Most agencies must stop immediately; the Pentagon has six months to remove Anthropic technology already embedded in military platforms. The clash came after Anthropic sought assurances that its Claude chatbot would not be used for mass surveillance of Americans or in fully autonomous weapons. Hegseth said the Pentagon “must have full, unrestricted access” for lawful purposes.

Hours later OpenAI announced a deal to deploy its models on the Pentagon’s classified network. Fortune reports that Sam Altman said it contains the same two limits Anthropic wanted, while OpenAI also says the Pentagon can use its technology for “any lawful purpose” — Fortune notes it is unclear how both are true. Pakistan’s The News reports that Anthropic said it was “deeply saddened” and plans legal action.

“We are asking the [Department of War] to offer these same terms to all AI companies.” — Sam Altman, via Fortune

How each outlet framed it

• AP — the standoff: Dario Amodei said Anthropic “cannot in good conscience accede”, and Altman said OpenAI shares the same red lines.
• Fortune — the business risk: the cancelled $200 million contract is small, but Hegseth’s bar on military contractors doing commercial business with Anthropic could put pressure on investors such as Amazon, Google and Nvidia.
• The News — the puzzle: Anthropic was the first lab on the classified network, and why the Pentagon accepted OpenAI’s limits is unclear.

What we don't know yet: whether the designation will hold. Fortune quotes legal experts who doubt it; Dean Ball called Hegseth’s broad reading “almost surely illegal”.