Three major US government agencies – the State Department, Treasury Department, and Federal Housing Finance Agency – are severing ties with artificial intelligence firm Anthropic this week, following an executive directive from President Donald Trump. The move comes as the Pentagon prepares to designate Anthropic a supply-chain risk, marking a dramatic reversal for the AI lab once considered vital to national security infrastructure.
In operational changes effective March 3, 2026, the State Department confirmed it will transition its internal StateChat system from Anthropic's Claude platform to OpenAI's GPT4.1 model. Treasury Secretary Scott Bessent and FHFA Director William Pulte separately announced via social media their departments’ full termination of Anthropic product usage, including mortgage agencies Fannie Mae and Freddie Mac.
The White House mandate allows a six-month transition period for defense applications, during which OpenAI has already secured a contract to deploy technology within the Pentagon's classified networks. This rapid realignment raises questions about the future of public-private AI partnerships, particularly for systems deemed critical to government operations.
Analysts suggest the policy shift could reshape the competitive landscape of defense-aligned AI development. While Anthropic faces unprecedented challenges in retaining federal contracts, OpenAI’s expanded government foothold positions it as the primary beneficiary of Washington’s revised tech procurement strategy.
