US Government Emergency Order Forces Anthropic to Shut Down Fable 5 and Mythos 5: Security Restrictions and Model Openness Conflict Intensifies

At 5:21 PM on June 12, 2026, the US Department of Commerce issued an emergency export control order to Anthropic, requiring it to shut down global access to Claude Fable 5 and Mythos 5. The order was based on researchers discovering a method to bypass Fable 5's security protections, which could threaten national security. Anthropic confirmed in a statement on its website that it had executed the shutdown, and access to all other models remains unaffected.

Model Core Capabilities and Practical Restrictions

Fable 5 and Mythos 5 were previously positioned as cutting-edge commercial models, serving hundreds of millions of users. After internal review, Anthropic believed the vulnerability produced only minor findings, and those findings could be replicated using publicly available models such as GPT-5.5. Anthropic explicitly opposed using a "narrow jailbreak" as grounds for recalling a deployed model, stating that if the industry uniformly adopted this standard, it would effectively halt all frontier model releases.

At the implementation level, API calls were interrupted immediately after the order was issued, and users could no longer access the two models from any region. Anthropic did not disclose the specific technical details of the vulnerability, only stating that its scope was smaller than the government's assessment.

Direct Comparison with Similar Products

Compared with OpenAI's GPT-5.5, Fable 5 showed a comparable level of risk in the same jailbreak tests. GPT-5.5 remains publicly available and has not been required to be taken down due to similar findings. Anthropic previously emphasized that its safety standards are above industry averages. This incident shows that the government has adopted stricter intervention standards for vulnerabilities that can be "replicated by other models."

Amazon had previously warned its clients about usage risks. The Department of Defense had earlier listed Anthropic as a supply chain risk and banned the use of the Claude series inside the Pentagon, a ban that has not been fully lifted to date.

Practical Impact on Developers and Businesses

Developers relying on Fable 5 or Mythos 5 for production deployments need to migrate models within 72 hours. Anthropic recommends switching to its unaffected models but has not provided migration tools or compatibility testing support. Enterprise users must reassess compliance risks, especially for applications involving sensitive data.

Within 48 hours of the order, the open-source community saw a surge in access, with some developers beginning to replicate and publicly disclose the relevant jailbreak methods.

Policy Enforcement and Business Consequences

Two days before the incident, Anthropic CEO Dario Amodei published an article arguing that the government should have the authority to block the release of models that do not meet safety standards, drawing an analogy to the Federal Aviation Administration's aircraft testing system. Sara Tortoli commented on Instagram that after a long history of describing models as potential civilization threats, it is not surprising that the government treats them as weapons.

For Anthropic's IPO process, the global access disruption could affect revenue forecasts and investor confidence. The ongoing lawsuit with the Trump administration has not been fully resolved, further increasing regulatory uncertainty.