The US government ordered on June 12, 2026 the immediate suspension of access to Fable 5 and Mythos 5, Anthropic's two most advanced AI models, for all foreign nationals. The directive, issued under national security authorities, comes just forty-eight hours after the commercial launch of Fable 5. What does this say about the regulatory trajectory of AI in the United States?
In brief
- On June 12, 2026, Washington imposed an export restriction suspending Fable 5 and Mythos 5 for all non-US citizens, including foreign employees of Anthropic.
- Anthropic disputes the decision and claims that the reported jailbreak technique only exposes a small number of already known minor vulnerabilities, also present in GPT-5.5.
- The other models in the Claude range remain operational; Anthropic says it is working to restore access as quickly as possible.
A directive 48 hours before the launch of Fable 5
Anthropic had introduced Fable 5 on June 10 as its most powerful publicly available model, designed for programming, finance and long-context tasks. Two days later, Washington stopped everything.
According to Anthropic, the government has not provided detailed evidence in support of its decision. The company argues, however, that allegations of jailbreaking on Fable 5 likely triggered the measure.
After examination, she claims that the technique in question reveals only a small number of minor vulnerabilities already documented, available in other competing systems.
Fable 5 had, however, been subjected to thousands of hours of testing by US government agencies, the UK AI Security Institute and private security teams. None of them had identified a universal method of circumventing the model's protections.
Mythos 5 in the viewfinder, when the Glasswing project complicates everything
Mythos 5, the other suspended model, is even more sensitive. Deployed as part of the Glasswing project in partnership with the US government, it is based on the same architecture as Fable 5 but operates with relaxed cybersecurity restrictions. Anthropic described it as having the most advanced cyber capabilities among AI systems available today.
The suspension is part of a broader context. The Trump administration has also reportedly extended restrictions on the export of advanced AI models to foreign governments, businesses and individuals, according to reports that have been circulating for several days.
The Mythos 5 case concentrates concerns: a third party would have succeeded in hacking the system, which would have directly fueled the government's decision.
Anthropic also warns that recalling commercial AI systems over limited safety concerns could set a dangerous precedent for the entire industry. A warning that takes on particular relevance as the company called, from June 5, for international coordination on advanced AI.
In short, the suspension of Fable 5 and Mythos 5 materializes in a few hours what Anthropic had been theorizing for weeks. States can now act quickly and strongly on AI models deemed strategic. A record launch, jailbreak allegations, a national directive, a global restriction.
Anthropic says it complies with the order while contesting it, and is working to restore access. The outcome of this standoff will set the tone for the entire industry. Stay tuned on Tremplin.io for further developments.
Maximize your Tremplin.io experience with our 'Read to Earn' program! For every article you read, earn points and access exclusive rewards. Sign up now and start earning benefits.
