The U.S. Department of Defense has stopped using Anthropic's AI tools, an official told the BBC1. Claude was in use as recently as last week, including in military operations against Iran, the BBC reported, citing sources.
The BBC's account places Claude inside an active operational context involving Iran, a claim that, if accurate, would mark one of the first publicly reported instances of a frontier large language model being used in live U.S. military operations against a named adversary. ANALYSIS The distinction between back-office analytical use and deployment tied to ongoing military operations carries different implications for how AI companies negotiate acceptable-use boundaries with defense customers.
Anthropic has published usage policies that restrict certain categories of harm.
The reported termination raises questions about the commercial relationship between Anthropic and the DOD. The BBC cited an official saying the department is no longer using Anthropic's tools, but the report did not specify whether the decision originated with the DOD, with Anthropic, or was mutual.
Separately, the same Techmeme snapshot notes that OpenAI CEO Sam Altman said there remains a fundamental difference in worldview between OpenAI and Anthropic on AI regulation, arguing that AI's benefits justify accepting some risks.
Australia's Joint Select Committee on Artificial Intelligence is scheduled to hear from both OpenAI and Anthropic this week as part of broader hearings on AI governance[1].