In an unprecedented move that has sent shockwaves through the global AI industry, the US government has forced Anthropic to suspend access to its two most advanced AI models, Fable 5 and Mythos 5, for all users worldwide. The directive, issued under national security authorities, bars any foreign national from accessing the models, a condition so broad that Anthropic determined it could only comply by taking the models offline entirely.
US Government Directive Targets Foreign National Access to Fable 5 and Mythos 5
Anthropic received the export control directive at 5:21 PM ET on June 12, citing national security concerns. The order explicitly prohibits access to both Fable 5 and Mythos 5 by any foreign national, whether they are located inside or outside the United States. This sweeping restriction even extends to Anthropic’s own employees who are foreign nationals, making operational compliance impossible without a global shutdown.
The company confirmed that the net effect of the order is that it must disable both models for all customers. Other Anthropic models, including Claude Opus 4.8, remain unaffected by the directive.
Timing and Impact of the Fable 5 Shutdown
The timing of the government order is particularly disruptive. Anthropic had only begun rolling out Fable 5 on June 9, offering it free of charge to all Pro, Max, and Enterprise customers through June 22. The model, which was handed to millions of users just three days before the directive, is now completely offline. Existing Fable 5 sessions end with an error, and new Platform requests to the model fail, forcing integrators to migrate to other models like Opus 4.8.
Fable 5 is the safeguarded version of Mythos 5. Both models share the same underlying architecture, but Fable incorporates safety guardrails that block or divert sensitive queries related to cybersecurity, biology, and chemistry. The unrestricted Mythos 5 is reserved exclusively for vetted government cyberdefenders and life sciences partners.
What Prompted the Government’s Action?
Anthropic’s assessment is that the order stems from a reported method to jailbreak Fable 5. The company reviewed a demonstration of the alleged bypass and determined it involved only minor, already-known bugs—the kind of vulnerabilities that other publicly available models can discover without any specialized exploit.
“To date, the government has only given us 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 stated. “Our understanding is that one potential jailbreak was shared with the government.”
The company expressed strong disagreement with the government’s rationale. “We are complying with the government’s legal directive and are removing access to Fable 5 and Mythos 5 for all users. However, 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. If this standard was applied across the industry, we believe it would essentially halt all new model deployments for all frontier model providers.”
Broader Implications for the AI Industry
Anthropic points out that the capability in question is widely available elsewhere, citing OpenAI’s GPT-5.5 as an example, and is used by cybersecurity defenders every day. The company maintains that the order is based on a misunderstanding and is actively working to restore access, promising more details within 24 hours.
The UK’s Minister for AI and Online Safety, Kanishka Narayan MP, weighed in on the situation, noting that the pause affects customers in both the US and UK. He framed the incident as a compelling case for technological sovereignty, pointing to the government’s £1.1 billion AI chip investment as a step toward reducing dependence on US-controlled AI infrastructure.
What Affected Users Should Do Now
For users and organizations that relied on Fable 5 or Mythos 5, the immediate priority is to migrate to alternative models. Anthropic has confirmed that new sessions will fall back to a user’s default model or Claude Opus 4.8. Integrators should update their API calls and workflows to use other available models to avoid service disruptions. Given the uncertainty around when—or if—access will be restored, organizations should also evaluate their reliance on any single AI provider and consider diversifying their AI toolchain to mitigate similar risks in the future. For those handling sensitive data, this incident underscores the importance of maintaining robust data backup and access control policies, as well as staying informed about regulatory changes that could impact the availability of critical AI services.
