Pentagon Got Help From Claude in Iran

Claude is still apparently welcome at the Pentagon, at least for now. The military used the AI tool from Anthropic in its attack on Iran over the weekend, according to separate reports in the Wall Street Journal and Axios. The move came hours after President Trump ordered federal agencies to begin phasing out their use of Anthropic's artificial-intelligence tools amid a showdown between the company and the government over how they can be deployed. The Pentagon, which used Claude in its operation to capture Nicolas Maduro, has the bot sift through intelligence, flag targets, and run war-game-style simulations, among other things, per the Journal.

The episode underscores how tightly woven Anthropic's tools have become in US military planning, even as the administration labels the company a security risk and moves to unwind the relationship over six months. In the meantime, the Pentagon is turning to rivals OpenAI and Elon Musk's xAI for classified work, amid concerns that the tools might eventually be used for mass surveillance. "We think our agreement has more guardrails than any previous agreement for classified AI deployments, including Anthropic's," said a statement from OpenAI CEO Sam Altman on Friday.