Commerce lifts export controls, Anthropic restores Fable 5

Anthropic restores – After weeks of negotiating with the Trump administration, Anthropic says the Department of Commerce has lifted export controls on Claude Fable 5 and Mythos 5. The company plans to begin restoring access Wednesday on Claude platforms worldwide, with AWS, Google

For days, Anthropic’s Claude Fable 5 sat sidelined—planned to be pulled back online, then suddenly not. Now the pause is ending.

Anthropic says it has received notice that the Department of Commerce has lifted export controls on Claude Fable 5 and Mythos 5. The company says it plans to begin restoring access Wednesday to users globally on Claude platforms. It also says access on AWS. Google Cloud. and Microsoft Foundry will be re-enabled soon. though it didn’t provide a specific timeline beyond that.

“ We’ll begin restoring access tomorrow, and will share an update soon,” Anthropic wrote in a post on X. The company said it is grateful to users for their patience and to everyone who worked with it on redeploying the models.

The return comes with a detailed internal reckoning. In a blog post published on Tuesday evening. Anthropic described the lead-up to the export-control decision. the safeguards it says it has revised. the new AI industry processes it says it is working on. and how it plans to share information with the government. The company also discussed prerelease testing for upcoming models.

In early June, Anthropic sidelined Fable 5. It framed Fable 5 as a consumer-facing model built from the same underlying technology as Mythos 5, but with more safeguards. The move followed a Friday evening ultimatum from the Trump administration. Anthropic says the government’s concern centered on potential jailbreaks of the technology. The export control directive. it reports. disallowed foreign nationals—from non-US members of enterprise client companies to even many of Anthropic’s own employees—from using either Mythos 5 or Fable 5.

The pressure had a name. Anthropic tied the export control directive’s motion to a jailbreak concern flagged by Amazon researchers, which it says was largely responsible for getting the directive put in place.

To address the jailbreak in question. Anthropic said it trained an improved safety classifier designed to target and block the behavior. The company wrote that users will be notified if a request to Fable 5 is blocked. and that the request will instead be sent to Opus 4.8. It also said the new classifier means the specific technique described in the Amazon report is blocked in over 99% of cases.

Just weeks earlier, the Trump administration had green-lit Mythos 5’s return—but only to a pre-approved list of organizations. Non-US members of those organizations. as well as Anthropic’s own foreign national employees. were allowed to regain access to Mythos 5. The company said that decision came shortly after OpenAI unveiled GPT-5.6. which the Trump administration allowed to debut under similar rules: a staggered rollout. initially limited to a pre-approved list of organizations and government departments.

On Tuesday, Anthropic said it will continue coordinating with the government to expand access to the broader set of domestic and international partners for Mythos 5.

The blog post also lays out how Anthropic says it intends to work going forward—especially in how models are tested and how problems are communicated. The company says it plans to offer “pre‑release government access and evaluation. ” particularly for models relevant to national security capabilities. In that arrangement. Anthropic says government partners can run independent evaluations about the models’ capabilities and test guardrails before any broader release. It also says the government will have access to Anthropic’s technical staff during those prerelease testing periods.

Anthropic said it would introduce “rapid information sharing” when “significant jailbreaks or misuse patterns are identified.” It also said it wants to work with the government and other leading AI labs to create a “shared, voluntary security and evaluation standard for frontier model providers.”

The company described operational commitments as well: it says it will “stand up dedicated Anthropic teams to work on shared government priorities. ” provide a significant compute allocation to support government testing and research. and make its safety and red‑teaming expertise available to help advance the state of the art in AI evaluation.

Behind these promises is a wider tension Anthropic described in its own words. It said there is currently no consensus in the AI industry for deciding on a jailbreak’s severity—and that the mismatch will become more acute as more models with powerful cybersecurity capabilities (and other capabilities) are trained. assessed. and released.

That concern runs through the company’s partnership plans. Anthropic said it partnered with Amazon. Microsoft. Google. and other enterprises in its Project Glasswing program to draft a widely-agreed-upon framework for assessing AI jailbreaks. It laid out four proposed categories: capability gain for the attacker. breadth of capability gain for the attacker. ease of weaponization more broadly. and discoverability—how easy it may be for someone else to repeat it.

Anthropic also said it created a new team to provide 24/7 monitoring of key jailbreak submission channels. It added that it will soon debut a HackerOne program for researchers to submit potential jailbreaks they have flagged for Fable 5.

In the blog post, the company included a blunt disclaimer. It wrote that “it is probably impossible to make any AI model fully robust (that is. impervious) to jailbreaks.” Anthropic said it expects some jailbreaks will be found for its models. with variations in severity. It said it expects many minor jailbreaks. some narrow harmful ones. and that while no universal jailbreaks for Fable 5 had been discovered at the time of writing. expert safety researchers would continue red-teaming the model.

The export-control shift arrives at a particularly sensitive moment for Anthropic. The company said the initial export control directive came at an inopportune time as it prepares for an IPO and while it has been feuding with the government for months over a supply chain risk designation.

Now, with the Department of Commerce lifting export controls on Claude Fable 5 and Mythos 5, the question for users is simple: how quickly can access be restored—and how quickly can the safeguards prove they’re the real answer to the jailbreak concerns that led to the shutdown in the first place.

Anthropic Claude Fable 5 export controls Department of Commerce Claude Mythos 5 AWS Google Cloud Microsoft Foundry jailbreaks AI safeguards