Fable 5 and Mythos 5 Return as US Lifts Anthropic Export Controls

Anthropic will restore global access to Claude Fable 5 and Mythos 5 on July 1, the company said Tuesday. The move ends a standoff that began when the US Commerce Department abruptly suspended the two models on June 12 over national security concerns. Commerce Secretary Howard Lutnick told Anthropic the controls were withdrawn after the company agreed to new obligations to detect security risks, collaborate on future model releases, and alert the government of malicious activity.

In a post on X, Anthropic said it had “received notice that the Department of Commerce has lifted export controls on Claude Fable 5 and Mythos 5” and would “begin restoring access tomorrow.” Fable 5 reaches Pro, Max, Team, and select Enterprise plans “for up to 50% of weekly usage limits through July 7, after which it will be available via usage credits,” the company said. Mythos 5 had already been partially unblocked on June 26, when the Commerce Department cleared it for more than 100 US institutions under the Glasswing program. Anthropic thanked users for their patience in a follow-up post.

Anthropic Says the Controls Are Lifted

The lifting covers both Fable 5, the consumer-facing model released on June 9, and Mythos 5, the version designed for select businesses and cybersecurity experts. A new safety classifier Anthropic trained with the Commerce Department now blocks the specific jailbreak technique that triggered the June 12 directive in over 99% of cases, the company said. Fable 5 will be re-enabled on AWS, Google Cloud, and Microsoft Foundry as quickly as possible, Anthropic added.

The Commerce Department’s Bureau of Industry and Security, in Lutnick’s letter to Anthropic on the conditions, said “a license is no longer required for the export, reexport, or in-country transfer, including deemed export or deemed reexport, of the Mythos or Fable models.” The agency reserved the right to reverse course, writing that it may “reconsider its decision to lift export restrictions if necessary.” Lutnick framed the deal as joint work in a post on X: “Over the past two weeks, we have worked closely with Anthropic to analyze and approve Fable 5 to ensure alignment across the U.S. Government and strengthen America’s leadership in AI.” He wrote that “Anthropic has agreed to proactively detect and address security risks associated with the models” and has also agreed to “collaborate on future releases of its AI models and alert the government of any malicious activity.”

OpenAI Hit the Same Wall a Day Earlier

One day before Anthropic’s announcement, OpenAI introduced its own next-generation family, GPT-5.6, with flagship Sol, balanced Terra, and faster Luna models, and disclosed that the Trump administration had asked it to limit initial access to a “small group of trusted partners.” Both companies are now subject to US government review of their frontier model releases.

OpenAI framed the limited rollout as a temporary compromise. “We are taking this short-term step because we believe it is the strongest path to broader availability in the coming weeks,” the company said in a blog post. Chief executive Sam Altman pushed back harder on X, writing that “this isn’t quite the process that we think is optimal,” per OpenAI’s statement on the limited GPT-5.6 rollout. OpenAI will work with the government “to attempt to get to a transparent, reliable process for early access,” he added.

OpenAI has not named the trusted partners. The model family is now restricted to those “whose participation has been shared with the government,” the company said. Sol costs $5 per million input tokens and $30 per million output tokens, Terra costs half that, and Luna costs $1 and $6.

Three Days of Public Access, Then a Friday Phone Call

The Anthropic saga began on June 9, when the company released Fable 5 and Mythos 5 globally, with Mythos 5 going only to vetted cybersecurity partners under the Glasswing program. Three days later, on June 12, the Commerce Department issued an export control directive at 5:21pm ET, citing a method of bypassing Fable 5’s safeguards that Amazon researchers had reported. The directive landed with no specific details of the underlying national security concern, Anthropic said. Anthropic disabled both models for all customers, including US-based foreign nationals, to ensure compliance with the directive.

Anthropic publicly disagreed with the move. In Anthropic’s original June 12 statement on the directive, the company said: “We disagree that the finding of a narrow potential jailbreak should be cause for recalling a commercial model deployed to hundreds of millions of people.”

1. June 9, 2026: Anthropic releases Claude Fable 5 and Mythos 5
2. June 12, 2026: Commerce Department issues export control directive at 5:21pm ET
3. June 12, 2026: Anthropic disables both models for all customers
4. June 26, 2026: Commerce allows limited release of Mythos 5 to 100+ US institutions
5. June 30, 2026: Commerce withdraws June 12 controls on Fable 5 and Mythos 5
6. July 1, 2026: Fable 5 returns globally

What Anthropic Changed to Come Back Online

Anthropic’s redeployment came with a specific technical fix. In a blog post Tuesday, the company said it had trained a new safety classifier that targets the jailbreak technique Amazon researchers had reported. Requests that would trigger the technique now route to Claude Opus 4.8, with a block notice shown to the user.

The new classifier was tested by researchers from the Commerce Department’s Center for AI Standards and Innovation (CAISI), who agreed the safeguards are “extraordinarily strong.” Anthropic acknowledged a trade-off: the new classifier “comes at the cost of flagging benign requests more often during routine coding and debugging tasks.”

Anthropic’s broader testing found that “many less capable models, including Claude Opus 4.8, GPT-5.5, and Kimi K2.7, could identify the same vulnerabilities as Fable 5 did in the report.” When it came to the demonstration of how to exploit the single vulnerability, every model Anthropic tested could produce the same code as Fable 5. The technique, the company said, reflected “a borderline case for Fable 5’s safeguards” rather than a Mythos-level capability.

The Commerce Department has cleared Mythos 5 for more than 100 US institutions under the Glasswing program, including major companies and government agencies, Semafor reported. Broader domestic and international access is still being negotiated.

A New Industry Framework for Jailbreaks

Anthropic is working with Amazon, Microsoft, Google, and other Glasswing partners to draft a consensus framework for rating the severity of AI jailbreaks. The framework would help AI developers triage new findings as they arise, the company wrote. It would also let them communicate the level of risk consistently to government and industry partners.

The point is to give developers, governments, and users a common yardstick at a moment when a single narrow finding can put a frontier model offline. A shared standard would also help labs launch highly-capable models with greater safety, the company wrote.

A common standard for assessing AI jailbreaks would help us and other companies launch new models safely, as well as allow our users to make the most of their advanced capabilities.

Anthropic made the case in its Tuesday blog post. The trigger for the June 12 directive was, in the company’s telling, a narrow finding the government had no shared framework to weigh. The original statement noted that the agency had only “verbal evidence of a potential narrow, non-universal jailbreak, which essentially consists of asking the model to read a specific codebase and fix any software flaws.” Anthropic argued a shared industry rubric would let researchers, labs, and regulators triage the same evidence in the same way.

The Limits Washington Put on the Deal

The export controls are gone, but Anthropic’s path back was paved with conditions. The Commerce Department has reserved the right to “reconsider its decision to lift export restrictions if necessary,” and Anthropic has committed to alert the government of malicious activity, collaborate on future releases, and accept deeper pre-release coordination.

The new obligations are reshaping how the most powerful models reach users. Dean Ball, a former White House AI adviser and soon-to-be OpenAI employee, said the Trump administration’s executive order, which asks AI companies to voluntarily submit their most advanced models for government review up to 30 days before release, has “created a de facto involuntary licensing regime for frontier AI, leading to heavy-handed restrictions.” Enforcement on the hardware side is moving in parallel: see the Taiwan prosecutors’ raid on Super Micro over Nvidia AI chip smuggling. Anthropic will turn Fable 5 back on for global users on Wednesday.

Frequently Asked Questions

When does Anthropic Fable 5 return?

July 1, 2026, is the date Anthropic has set for Fable 5’s global return. The rollout will hit the Claude Platform, Claude.ai, Claude Code, and Claude Cowork, with a cap of 50% of weekly usage limits on Pro, Max, Team, and select Enterprise plans lasting through July 7.

Why were export controls placed on Fable 5 and Mythos 5?

On June 12, 2026, at 5:21pm ET, the Commerce Department ordered Anthropic to suspend access to Fable 5 and Mythos 5 for any foreign national, whether inside or outside the United States. The company disabled both models for all users to ensure compliance, since it had no reliable way to verify nationality in real time.

What did Anthropic change to bring the models back online?

Anthropic trained a new safety classifier that specifically targets the Amazon-reported jailbreak technique. The Commerce Department’s Center for AI Standards and Innovation tested the safeguards and said they were extraordinarily strong. Requests that would have triggered the technique now route to Claude Opus 4.8.

Is Mythos 5 also returning for everyone?

No, not yet. The June 26 partial approval cleared Mythos 5 for more than 100 US institutions, including major companies and government agencies. Broader international availability is still being negotiated under Anthropic’s Glasswing program.

Does this affect OpenAI’s GPT-5.6 release?

OpenAI released GPT-5.6 Sol, Terra, and Luna on Friday and said the Trump administration had asked the company to limit initial access to a “small group of trusted partners.” Sam Altman wrote on X that OpenAI will work with the government “to attempt to get to a transparent, reliable process for early access.”