US lifts export controls on Anthropic’s Fable 5, clearing its return

The US Commerce Department has reversed the export controls it imposed on Anthropic’s most advanced AI models, ending a roughly three-week freeze that had forced the company to take its Claude Fable 5 system offline. The decision, first reported by Reuters on Tuesday, 30 June, was confirmed by Anthropic in a blog post titled “Redeploying Claude Fable 5” just hours later.

The restrictions had covered both Fable 5, the consumer-facing model, and Mythos 5, the more powerful cybersecurity system underpinning it. Both were cleared simultaneously. Commerce Secretary Howard Lutnick announced that an export licence would no longer be required. In exchange, Anthropic agreed to proactively detect and address security risks, help develop standards for future models, and report malicious activity to the government.

The original controls were imposed around 12 June, forcing Anthropic to shut down both models almost overnight. The trigger was a jailbreak discovered by researchers, reportedly at Amazon, who found a way to bypass Fable 5’s guardrails and reach the underlying cyber capabilities of Mythos. Anthropic disputed the severity of the vulnerability throughout the episode, arguing the flaw did not warrant a full shutdown. Regulators took a harder line, at least initially.

The distinction between the two systems is crucial here. Mythos 5 is the powerful cybersecurity model, and Fable 5 is the consumer-facing layer built on top of it. That architecture is precisely why a jailbreak of one raised alarms about the other. The original clampdown may have been tied to concerns that a China-linked group could gain access to Mythos through the exploit, though that motive rests on anonymous sourcing and should be treated as reported rather than confirmed.

The government’s position appears to have softened over the following weeks. Several outlets have framed the reversal as a response to industry criticism and worries that rivals abroad could gain ground while a leading US model sat frozen, though that framing is analysis rather than official explanation.

What comes back, and when? Anthropic said Fable 5 will return globally on 1 July across its products, including the Claude platform, Claude.ai, Claude Code, and Claude Cowork. The rollout is deliberately gradual. Usage will be capped at up to 50% of normal weekly limits through 7 July before returning to full availability, according to the company’s post.

The three-week saga has been unusually public for an export-control dispute, most of which play out quietly through the Bureau of Industry and Security. Here, the model was named and the company narrated its own suspension and revival in near real time. It also sits inside a broader tightening of US policy on where advanced AI can travel. Washington has spent the past year closing loopholes that let cutting-edge chips and models reach Chinese firms through overseas subsidiaries.

For Anthropic, the immediate stakes were commercial as much as regulatory. A frozen flagship model means paying customers routed elsewhere and a competitive gap that widens by the day. The company had spent much of June in and out of Washington, meeting officials to argue its case as the freeze dragged on. The conditions Lutnick attached to the reversal read as the price of getting back online.

Anthropic had already been cleared to restore Mythos 5 for a set of trusted cyber-defence partners before this week, while Fable 5 stayed restricted. Tuesday’s decision closes that remaining gap.

What is on the record is narrow and clear. Commerce lifted the controls, Lutnick set conditions, Anthropic accepted them, and access is returning. What remains hazier is the politics behind the reversal, from the exact role of the alleged China link to how heavily competitive pressure weighed on the final call. On those points, the sourcing is thinner, and the fuller account will likely emerge over the coming weeks.