Anthropic Pulls Mythos, Fable Models as US Bans Foreign Access

Anthropic said it disabled access to its Fable 5 and Mythos 5 artificial-intelligence models after receiving a U.S. export-control directive tied to national-security authorities.

According to the company, the order required it to suspend access to those two models for foreign nationals, whether they were inside or outside the United States, and it explicitly included foreign-national employees. Anthropic said it then shut both models off for all customers to ensure compliance. The company added that its other models remain available.

That leaves a narrow but immediate service impact in confirmed terms: customers using the two affected models face an interruption, while customers relying on the rest of Anthropic’s lineup do not, based on the company’s statement.

Anthropic has not publicly said how long the shutdown will last or whether it expects to restore access for permitted users once it has a more targeted compliance process in place. It also has not described whether any customers received advance warning before the cutoff.

For businesses that built products or workflows around the unavailable systems, the practical issue is not the wider platform but the loss of those specific model endpoints.

Analysis: The episode shows how a targeted government restriction can produce a broader service outage when the legal rule is narrower than the fastest workable technical response.

A nationality-based access rule is harder to apply than a simple country block because it turns on who the user is, not just where the user connects from, and that can complicate enforcement across global companies, shared workspaces and cloud-based tools.

Anthropic has not publicly detailed the mechanics behind its decision to disable the two models for everyone, beyond saying the broader shutdown was necessary to comply. Without that detail, it is possible to say only that the company chose the more sweeping service response rather than attempt immediate user-by-user filtering.

The directive’s inclusion of foreign-national employees may also have internal implications, although those are not public. If certain staff members cannot access the affected models, that could complicate research, testing, support or compliance workflows tied to them.

Anthropic has not described any such changes, so any effect inside the company remains a possibility rather than an established consequence.

The broader market significance is less about the number of models affected than about the speed with which access conditions can change for advanced AI systems. For developers and corporate buyers, the incident is a reminder that frontier-model availability can be shaped not only by price, performance and computing capacity,

but also by export controls and national-security policy. That adds another layer of planning for companies that depend on a specific model for production use, especially if alternatives are not interchangeable.

The next steps are still uncertain because key details have not been disclosed. It is unclear whether the shutdown is expected to be brief while Anthropic builds a narrower screening mechanism, or whether the restriction could persist for longer.

For now, the confirmed picture is straightforward: two Anthropic models were disabled after a U.S. directive restricting foreign-national access, the company applied that restriction by turning those models off more broadly, and customers using them are dealing with the resulting disruption.

Published at 2026-06-13T04:00:51.591912+00:00 UTC

Related Symbols

• STT — State Street Corporation
• Selection note: Fallback selected only high-confidence company matches from the candidate set.

References

• Bloomberg
• CNBC