Anthropic AI Model Suspension Shocks Developers and Foreign Users

Anthropic AI Model Suspension Shocks Developers and Foreign Users

Quick Reads

• The US government issued an export control directive ordering Anthropic to suspend access to Fable 5 and Mythos 5.
• The order bars all foreign nationals from the models, including Anthropic’s own international employees.
• Anthropic received the directive on Friday at 5:21pm ET and cut off access globally shortly after.
• Anthropic disputes the government’s rationale and says the identified vulnerability also exists in rival models.
• The move has reignited debate over AI sovereignty and export controls in Asia and beyond.

Anthropic AI model suspension caught developers and users off guard last Friday, when the company abruptly pulled access to two of its newest models. The US government issued an export control directive ordering Anthropic to block all foreign nationals from Fable 5 and Mythos 5, citing national security concerns. Anthropic received the order at 5:21pm ET and moved fast to comply.

Consequently, the company shut down both models for all users globally. This included foreigners physically present in the United States. Notably, it also affected foreign national employees working inside Anthropic itself.

In a blog post published Friday, the company said the directive did not explain its specific security concern in detail. Nevertheless, Anthropic acted swiftly, choosing compliance over a grace period or negotiation.

The Mythos AI model sits at the heart of this controversy. It is particularly adept at detecting software vulnerabilities, some of which have gone undiscovered for decades. This capability has previously aided US authorities and selected companies in strengthening their systems. However, the same capability raised fears about what could happen in the wrong hands.

Furthermore, Fable 5, released just days before the order, is a Mythos-class model built for autonomous knowledge work. Anthropic had already implemented additional guardrails on Fable 5, blocking its advanced cybersecurity and biotechnology capabilities. Additionally, Mythos 5, the non-public full version, was intended only for government agencies and select corporate partners.

Anthropic pushes back on the government’s reasoning. The company said it reviewed a report likely behind the directive and concluded it referred to a narrow capability to review and correct specific programme code. Moreover, Anthropic stressed that rival models, including GPT-5.5 from OpenAI, carry the same capability.

“Software used by hundreds of millions of users should not be blocked for this reason,” the company argued, citing extensive safety testing on Fable 5.

The suspension also deepens an ongoing feud between Anthropic and the Trump administration. Earlier in the year, Anthropic declined to allow the Pentagon to deploy its models for domestic surveillance and autonomous weapons systems. In response, the Department of Defense threatened to classify Anthropic as a supply chain risk.

Meanwhile, the fallout has spread globally. South Korea, which was deepening AI partnerships with Anthropic, now finds itself locked out of the company’s most powerful tools. Analysts say the move underscores a broader push for AI sovereignty across Asia and other regions. Beyond that, Anthropic’s other models, including Opus 4.8, remain active and unaffected by the directive.