Bold take: the White House questions whether a deal between Anthropic and the Pentagon can survive the current turmoil.
But here’s where it gets controversial: Anthropic’s CEO, Dario Amodei, reportedly disparaged the Trump administration in internal notes, a move that could derail talks and complicate any path to reconciliation with the Pentagon, according to an administration official cited by Axios.
Why this matters: With rival OpenAI and lawmakers from both parties pushing for an agreement, the Pentagon’s threat to label Anthropic a supply-chain risk adds pressure to reach a deal.
What’s happening: The Information reported that Amodei sent a memo to employees last Friday claiming President Trump disliked Anthropic for not offering him “dictator-style praise.” He allegedly described the OpenAI-Pentagon arrangement as “safety theater,” per the report.
What officials are saying: An administration official emphasized that the central goal is ensuring frontline soldiers have the best tools, and questioned whether Claude could secretly align with Amodei’s broader agenda in a classified context. Anthropic has insisted it does not seek operational control over how the Pentagon uses Claude, arguing that responsibility should rest with warfighters. People familiar with the matter also noted that Anthropic executives told the Pentagon they regret the media coverage didn’t adequately reflect their stance.
The other side: In recent days, both sides had been making progress toward a settlement, and it isn’t clear how much the leaked memo will derail negotiations.
Bottom line: This has been a highly visible clash between the Pentagon and Anthropic, marked by sharp exchanges. At this point, the basic dynamics appear unchanged.
