Anthropic vs. The U.S. Government: The Full Timeline

Why it matters

Anthropic's standoff with Washington set a template for how governments can pressure frontier AI companies: through contracts, blacklists, lawsuits and export controls.

Dario Amodei's confrontation with Washington is now the subject of Red Lines, a RuntimeWire documentary premiering today that reconstructs how Anthropic's dispute with the U.S. government moved from contract language to a federal blacklist, two court fights and the first export-control takedown of a commercial frontier AI model.

https://www.youtube.com/watch?v=vEY2UpP9KZQ

The film covers the four-month sequence that began with a Pentagon ultimatum in late February and ended, at least for now, with Anthropic restoring access to Claude Fable 5 and Claude Mythos 5 after the Commerce Department lifted export controls on June 30. The story turns on a narrow refusal by Amodei, Anthropic's co-founder and CEO: he would keep Claude's restrictions against mass domestic surveillance and fully autonomous weapons, even if the Pentagon wanted contract language allowing use of the system for any lawful purpose.

Amodei's position was not a clean break with national-security work. In his Feb. 26 statement, he wrote that Anthropic had worked to deploy its models to the Department of War and the intelligence community, and that partially autonomous weapons already used in Ukraine are important to democratic defense. His objection was aimed at two uses Anthropic said current AI systems should not support: AI-driven mass surveillance and weapons that remove human responsibility from lethal force. "We cannot in good conscience accede to their request," Amodei wrote in that statement.

That sentence became the dividing line. On Feb. 27, President Donald Trump ordered federal agencies to stop using Anthropic technology, according to contemporaneous reports and later court filings. Defense Secretary Pete Hegseth then designated Anthropic a "Supply-Chain Risk to National Security," a label that threatened to push Claude out of military contracting and force defense suppliers to reconsider their own Anthropic usage.

Anthropic answered in court on March 9. The company sued in the Northern District of California over the presidential directive, Hegseth directive and Department of War supply-chain designation. It also filed a separate petition in the D.C. Circuit challenging a broader procurement designation. The dual-track strategy mattered because the government had used overlapping authorities, and each track put a different part of the blacklist before a different court.

The first major ruling came March 26. Judge Rita F. Lin issued a preliminary injunction blocking the federal-agency ban and the Department of War's supply-chain-risk designation while the case proceeded. The order did not force the Department of War to keep using Anthropic. It barred the government from enforcing the specific directives that had followed the February standoff.

The D.C. Circuit took a different procedural posture on April 8. It denied Anthropic's request for a stay of the broader procurement designation while the appeal moved forward, leaving part of the government's supply-chain action in force. Legal analysts described the split as a function of two statutes and two standards of review rather than a direct contradiction between the courts. For Anthropic customers and contractors, the result was more complicated: parts of the ban were blocked, while a separate procurement designation still mattered inside defense contracting.

Then the fight moved from procurement to model access. On June 9, Anthropic launched Claude Fable 5 and Claude Mythos 5. Anthropic described Fable 5 as a Mythos-class model made safe for general use, with safeguards that route some risky requests to Claude Opus 4.8. Mythos 5 used the same underlying model with some safeguards lifted for a smaller set of trusted cybersecurity partners in Project Glasswing.

Three days later, Anthropic took both models down. In a June 12 statement, Anthropic said the U.S. government had issued an export-control directive suspending access to Fable 5 and Mythos 5 by any foreign national, including non-U.S. Anthropic employees. Because the order applied immediately and Anthropic said it had no reliable way to verify nationality in real time, the company suspended access for all users.

That decision produced the documentary's central commercial fact: for 18 days, Anthropic's most capable model line was unavailable worldwide. The company had launched Fable 5 at $10 per million input tokens and $50 per million output tokens, less than half the price it had set for Claude Mythos Preview. The shutdown therefore hit at the exact moment Anthropic was trying to widen access to the capabilities it had previously limited to a small trusted group.

Anthropic's June 30 redeployment post gives the clearest account of the technical trigger. The company said the export-control directive followed a report from Amazon researchers who found a way to bypass Fable 5 safeguards so the model identified software vulnerabilities and, in one case, produced code demonstrating how a vulnerability could be exploited. Anthropic said its own testing found the behavior did not show unique Mythos-level cyber capability because other models, including older Claude models and non-Anthropic models, could produce the same exploit demonstration.

The government lifted the export controls on June 30. Anthropic restored Fable 5 globally on July 1 across Claude Platform, Claude.ai, Claude Code and Claude Cowork, with cloud-provider access to follow. Mythos 5 returned on a narrower basis for approved U.S. organizations, with broader Project Glasswing access still dependent on government coordination.

The settlement around model access did not erase the precedent. Anthropic agreed to more government pre-release evaluation, faster information sharing around major jailbreaks and work with Amazon, Microsoft, Google and other Glasswing partners on a shared framework for rating cyber jailbreak severity. That is the new bargain at the center of the film: a model developer that built its brand around safety red lines kept the red lines that started the Washington fight, while accepting a deeper government role in testing the next models before release.

For founders building in AI, the Anthropic case now reads less like a one-off fight over Pentagon procurement and more like a map of the pressure points around frontier models. Contract terms, product safeguards, export controls, cloud distribution, citizenship checks, partner access and courtroom jurisdiction all became levers in the same dispute. Red Lines follows those levers one by one, from Amodei's February refusal to the June takedown that showed how quickly a policy fight can become a global product outage.