US Order Forces Anthropic to Suspend Fable 5 and Mythos 5 Worldwide

Anthropic has suspended global access to its Fable 5 and Mythos 5 AI models after receiving a US government export control directive citing national security concerns, marking an unprecedented intervention in the deployment of a frontier AI model.

The company said the order requires it to block access to both models for all foreign nationals, whether inside or outside the United States, including Anthropic's own non-US employees. To comply with the directive, Anthropic has disabled the models for all customers worldwide.

The company said the order was issued under national security authorities but did not specify the underlying concerns.

According to Anthropic, the government believes it has identified a method of "jailbreaking" Fable 5—a technique used to bypass a model's built-in safety controls.

However, Anthropic disputed the severity of the finding, arguing that the reported jailbreak only uncovered a small number of previously known software vulnerabilities and did not provide capabilities beyond those already available from other leading AI models.

The company said its internal review found that the demonstrated technique could identify only minor flaws and that similar results can be achieved using publicly available models, including OpenAI's GPT-5.5.

Anthropic emphasized that Fable 5 underwent extensive safety testing before launch, including thousands of hours of red-team exercises conducted with US government agencies, the UK's AI Safety Institute and independent third-party organizations.

According to the company, those evaluations found Fable's cybersecurity safeguards to be stronger than those of previously deployed frontier models and failed to identify a universal jailbreak capable of broadly bypassing its protections.

Instead, Anthropic said it adopted a "defense in depth" approach that combines layered safeguards with monitoring systems designed to detect and mitigate misuse quickly.

The company noted that it introduced a 30-day customer data retention policy specifically to help identify and respond to jailbreak attempts, despite acknowledging that the policy carried commercial costs.

While complying with the government's directive, Anthropic publicly challenged the decision, arguing that recalling a commercial AI model over what it described as a narrow, non-universal jailbreak sets a problematic precedent for the industry.

"If this standard was applied across the industry, we believe it would essentially halt all new model deployments for all frontier model providers," the company said.

Anthropic added that it supports government oversight of advanced AI systems but believes intervention should occur through a transparent statutory process grounded in technical evidence rather than ad hoc directives.

The move highlights growing tensions between rapid AI innovation and national security oversight as frontier models become increasingly capable of software engineering and cybersecurity tasks.

For enterprises, the suspension underscores a new operational risk associated with relying on frontier AI services. Beyond technical performance and pricing, organizations may now need to account for regulatory interventions that could abruptly restrict access to critical AI models used in software development, security operations and business workflows.

The incident is likely to intensify debate over how governments should regulate advanced AI capabilities while balancing innovation, security and commercial continuity.