For months, the conflict between Anthropic and the Pentagon seemed to center on access to Claude. Court documents released this week indicate it was about something larger: who controls how the US military applies frontier AI.

The emails appeared on Tuesday in one of the lawsuits Anthropic filed against the Department of Defense. The Wall Street Journal first reported them. They detail a tense exchange between two individuals. One is Anthropic chief executive Dario Amodei. The other is Emil Michael, the undersecretary of defence for research and engineering.

The redline

Amodei maintained a steady stance. He sought clear limits on Pentagon use of Anthropic models. Two applications were ruled out: fully autonomous weapons and domestic surveillance. Those match the boundaries he outlined when questioning military applications of his models.

The Pentagon sought wider scope. It requested coverage for all lawful uses, a broad phrase that allows significant flexibility.

The of EU tech

The latest rumblings from the EU tech scene, a story from our wise ol' founder Boris, and some questionable AI art. It is free, every week, in your inbox. Sign up now!

Discussions deteriorated in January. Michael emailed Amodei after weeks without reply. He noted he was hoping the sides were closer to engaging with the revised position. In clear terms, he wanted Anthropic to move toward the Pentagon view. Amodei restated his redlines instead.

Just not workable

Michael was direct. The guardrails were just not workable, he wrote. He gave Anthropic one more chance to align on core principles before the sides parted ways. He also dismissed the boundary Amodei set. There is no distinction in our world between weapons that are defensive or offensive, he wrote. Gizmodo published the emails from the court record.

Amodei challenged the all lawful uses standard. The issue, he noted, is that US law does allow domestic surveillance. He told Michael the Pentagon draft language went too far. It appeared to completely remove our redlines.

The next day, Defence Secretary Pete Hegseth labeled Anthropic a supply-chain risk. That designation usually targets firms linked to foreign adversaries. It converted a months-long standoff into a court battle.

A conflicted negotiator

Michael role has attracted attention. Financial disclosures show he held stock in xAI, an Anthropic rival, along with other AI investments. The Guardian and ProPublica reported the holdings. Anthropic claims the blacklisting was retaliation rather than security policy. A federal judge agreed in late March. She granted a preliminary injunction.

She described the action as classic illegal First Amendment retaliation. An appeals court reversed that in April, and the case continues. The dispute forms one thread in Silicon Valley growing connections to the Pentagon.

Why Europe should watch

The stakes extend beyond Washington. Anthropic is testing whether an AI firm can impose ethical limits on a government customer while retaining contracts. That question lies at the heart of Europe own debates over military and surveillance AI. The EU AI Act is addressing the same boundaries. It also fuels the sovereignty discussion, as European buyers consider how much control any US lab retains once its models enter national-security work.

For now, Anthropic has stabilized its position. It resolved its latest clash with the administration this week, moving toward a deal to restore access to a restricted model. The emails illustrate how near the relationship came to collapse, and why.