US Defence Secretary Pete Hegseth has summoned Dario Amodei, CEO of AI firm Anthropic, to the Pentagon for a critical meeting focused on the US military’s use of the company’s advanced artificial intelligence system, Claude. Officials and media reports describe the gathering as a potentially “tense confrontation” over terms for expanding access to Claude within classified military networks.

The Pentagon is pushing AI developers to make their most capable models available with fewer restrictions for use in defence and intelligence workflows. Claude is currently the only frontier AI model deployed in certain classified environments and is considered highly capable for complex analytical tasks. However, Anthropic has resisted fully lifting guardrails governing how Claude can be used, especially regarding mass surveillance or autonomous weaponisation.

Tensions escalated after reports that Claude was used during a recent US military operation in Venezuela, drawing scrutiny over its operational role. The Defence Department has also warned it may designate Anthropic a “supply chain risk” if the company does not agree to the Pentagon’s terms — a designation that could sever its military ties and force partners to avoid using Claude.

The outcome of the meeting may have significant implications for how cutting-edge AI technologies are governed and deployed within national security frameworks.