US Government Suspends Access to Anthropic's Fable 5 and Mythos 5 AI Models

The US government, citing national security authorities, has issued an export control directive to suspend all access to Anthropic's advanced artificial intelligence models, Fable 5 and Mythos 5, for all its customers, including foreign national employees, effective June 12, 2026.

This directive, received by Anthropic on June 12, 2026, at 5:21pm (ET), mandates the abrupt disabling of these models for all users. Notably, the government's letter did not provide specific details regarding the national security concern driving this action.

Anthropic, for its part, maintains that its safeguards are substantially more effective than those of any previously deployed model, and that no "universal jailbreak" has yet been found.

Anthropic, for its part, maintains that its safeguards are substantially more effective than those of any previously deployed model, and that no "universal jailbreak" has yet been found.

Anthropic's understanding is that the government believes it has become aware of a method for bypassing, or "jailbreaking," Fable 5. However, the company reviewed a demonstration of this specific technique and found it only identified a small number of previously known, minor vulnerabilities. These vulnerabilities, according to Anthropic, appear relatively simple and can also be discovered by other publicly available AI models without requiring a bypass.

The company has implemented strong safeguards for Fable 5, significantly reducing the likelihood of misuse for tasks related to cybersecurity. In fact, these protections are so robust that many users have complained they are overly broad.

In the weeks leading up to Fable 5's launch, Anthropic collaborated with the US government, the UK AISI, multiple private third-party organizations, and internal teams, dedicating thousands of hours to red-teaming Fable's safeguards. These extensive tests showed that Fable's safeguards were substantially more effective than those of any previously deployed model.

Anthropic acknowledges that perfect jailbreak resistance is not currently possible for any model provider, as every safeguard in the industry is vulnerable to "non-universal jailbreaks" that can elicit some cyber information in specific circumstances. The company had clearly stated this when Fable 5 was released.

Given this reality, Anthropic adopted a "defense in depth" strategy for Fable 5. The goal was to make jailbreaks either narrow or very expensive to produce, combining this with thorough monitoring to quickly detect and shut down any successful attacks. This is also why Anthropic required 30-day retention of customer data for Fable, a policy change that carried real costs for the company and its customers but allowed for research and mitigation of jailbreaks.

The company stands by its defense in depth strategy, asserting that it reduces Fable's risks to a level comparable to those of existing models already deployed across the industry. To date, Anthropic has not received disclosure of any concerning non-universal potential jailbreak that led to a harmful result, with reported potential jailbreaks being either benign or minor findings without significant impact.

The government has only provided verbal evidence of a potential narrow, non-universal jailbreak, which essentially involves asking the model to read a specific codebase and fix software flaws. Anthropic validated that the capability level displayed in the report, which they believe forms the basis of the government's directive, is widely available from other models like OpenAI’s GPT-5.5 and is used daily by cybersecurity defenders to keep systems safe.

While Anthropic is complying with the legal directive, the company disagrees that the finding of a narrow potential jailbreak should warrant recalling a commercial model deployed to hundreds of millions of people. If this standard were applied across the industry, Anthropic believes it would essentially halt all new model deployments for all frontier model providers.

This abrupt decision raises critical questions about the balance between national security and AI innovation, especially as Fable 5 was already being integrated by partners like Tata Consultancy Services (TCS) and DXC Technology to serve regulated industries globally.