Pentagon Cuts Ties with Anthropic Over AI Use Dispute

Newly revealed US court documents expose a weeks-long conflict between AI company Anthropic and the Pentagon, centered on the ethical use of Anthropic's AI model Claude. The dispute began in early January when the Pentagon demanded unrestricted use of Claude, including for surveillance and autonomous weapons systems. Anthropic CEO Dario Amodei pushed back, seeking clear limits on using AI for civilian surveillance and weaponry, emphasizing that legality alone did not justify such uses.

On February 4, Emil Michael, Deputy Secretary of Defense for Research and Engineering, rejected Anthropic's restrictions, insisting on full operational freedom for any lawful AI application. Amodei acknowledged the growing impasse, and despite further compromise attempts by Anthropic, the Pentagon dismissed them as inadequate. By February 27, Amodei conceded that cooperation was no longer possible.

The following day, Defense Secretary Pete Goss directed that Anthropic be designated a "supply chain risk," barring Pentagon contractors from using its AI models. President Donald Trump also ordered government agencies to cease using Anthropic's tools, though a federal judge temporarily blocked some restrictions pending legal review. Meanwhile, federal agencies scrambled to remove Claude from their systems, with about two-thirds of Pentagon units already transitioning to alternative platforms.

Anthropic responded by stating the legal battle was necessary to protect the company and its partners, while reaffirming its commitment to working productively with the government to ensure safe and reliable AI benefits for all Americans.