Sources: US Defense Secretary Pete Hegseth summons Dario Amodei to the Pentagon on Tuesday morning for what is likely to be a tense meeting over Claude usage
AI can't kill humans without a man in the loop, and no mass surveillance of Americans — are 100% consistent with our military's values and pre-existing rules, and with the views of most people across partisan lines. Where the hell is Congress here?Danielle Fong /@daniellefong:claude prepare me for this meeting. make no mistakesJason Liu /@jxnlco:Let's see if Dario passes the cultural fit interview questions@jmbollenbacher:i feel like dario will fold here But if he doesnt fold and the Trump admin
The meeting became the opening point of a longer negotiation: Anthropic was later reported to be holding contract talks with the Defense Department, while maintaining that it would not relax Claude’s military restrictions.
First-order effects
Anthropic and Dario Amodei face immediate Pentagon pressure to clarify or alter the conditions under which Claude can be used by the military.
The Pentagon must determine whether Claude can be deployed under Anthropic’s existing safeguards or whether access will require a revised agreement.
Second-order effects
Contract terms—not just model capability—become a procurement variable for military AI adoption, giving safety restrictions direct commercial and operational weight.
Other frontier-model providers seeking Pentagon work may face sharper scrutiny over whether their usage policies align with defense requirements.
Third-order effects
If similar disputes persist, the defense AI market may separate into labs willing to operate under government-defined access rules and labs that retain independently set safeguards.
The episode reinforces dual-use AI governance as a contest over who sets deployment boundaries: model developers, defense buyers, or Congress.
The trend: Frontier AI labs are increasingly being judged on their willingness to make model-access policies compatible with national-security procurement.
This makes it very clear: Hegseth and the Trump administration want killer robots and AI mass surveillance of Americans. Anthropic doesn't want this. Implication is that OpenAI, Google, and xAI are okay with it. [image]
Anthropic's conditions — AI can't kill humans without a man in the loop, and no mass surveillance of Americans — are 100% consistent with our military's values and pre-existing rules, and with the views of most people across partisan lines. Where the hell is Congress here?
i feel like dario will fold here But if he doesnt fold and the Trump admin actually exiles Anthropic from DoD work, thats actually amazing news Anthropic leaving DoD behind would reduce AI s-risk odds tremendously. It'd also help Claude, and thatd help Anthropic in the long run
Dario Amodei is being summoned by the Department of War, he will meet with Pete Hegseth tomorrow morning. The Axios report quotes a Defense official who claims Hegseth will be presenting Amodei with an ultimatum. [image]
Via @Axios Scoop: Hegseth to meet Anthropic CEO and issue ultimatum as Pentagon threatens banishment. BUT “Claude is the only AI model available in the military's classified systems, and the most capable model for sensitive defense and intelligence work. The Pentagon doesn't want
Axios: Secretary Hegseth has summoned Anthropic CEO Dario Amodei to the Pentagon on Tuesday morning for what sources say is likely to be a tense meeting over terms for military use of Claude. “This is a shit-or-get-off-the-pot meeting,” a senior Defense official said.
Big Brother wants to use Ai for mass surveillance, and — making autonomous weapons of war. We can't let any — Ai company sell it's product to use against people. — www.axios.com/2026/02/23/h... [image]
Anthropic's models are so good, the U.S. military is threatening them into letting their AI be used to create weapons and for global surveillance. — This is ironically a great sales pitch for Claude Code.