For about three weeks in June 2026, two of Anthropic’s most advanced AI models simply went dark for most users. The Trump administration had imposed export controls that forced the company to cut off access industry-wide. Now, after negotiations that reached the highest levels of the Commerce Department, those restrictions are fully lifted.

What happened, and how it got resolved

On June 12, 2026, the Trump administration issued export controls targeting Anthropic’s two flagship models, Mythos 5 and Fable 5. The Department of Commerce cited national security concerns, specifically the risk that adversaries could find ways around safeguards protecting sensitive cybersecurity capabilities built into the models.

Anthropic had no real choice but to comply. The company disabled both models broadly, affecting customers across industries who had built workflows around them.

Commerce Secretary Howard Lutnick became the central figure in the negotiations that followed. On June 26, Lutnick authorized a partial restoration, allowing more than 100 trusted U.S. organizations to regain access to Mythos 5, the more powerful of the two models. Fable 5 access followed. By July 1, the administration had fully reversed the export controls on both.

Anthropic CEO Dario Amodei and co-founder Tom Brown were directly involved in communications with the administration throughout the process, according to reporting from the New York Times.

Mythos 5, even post-reinstatement, carries a distinction: it remains designated primarily for trusted partners operating within critical infrastructure sectors.

Why this matters beyond one company’s models

Anthropic arriving at this moment is not entirely surprising. The company had previously faced friction with the Defense Department, including a period on a blacklist that complicated its federal contracting work. The June export control action represented an escalation of that tension into commercial product access.

Export controls were issued June 12. A partial reversal came June 26. Full restoration followed July 1. That is less than three weeks from shutdown to reinstatement, which is unusually fast for a regulatory action of this type.

What investors and the crypto market should watch

Advanced AI models are increasingly being deployed for on-chain security, fraud detection, and transaction monitoring across blockchain networks. If access to the most capable models is subject to government restriction, projects relying on those models for infrastructure security face a new category of operational risk.

The restoration of access is positive for that ecosystem, at least in the short term. But the episode demonstrated that the risk is real, not theoretical.

For investors watching the AI sector, Anthropic’s models were effectively off the market for paying customers for nearly three weeks. The company had no ability to prevent that.