Anthropic suspends Fable 5 and Mythos 5 access following US government directive
Company disagrees with the move, arguing the recall lacks transparency and technical grounding despite a narrow reported vulnerability.
Anthropic has suspended all customer access to its newly launched artificial intelligence models, Fable 5 and Mythos 5, to comply with an immediate directive from the US government. The order, received at 5:21pm ET on Friday, 12 June, cites national security concerns and mandates the cessation of access for all foreign nationals, regardless of their location or employment status with the company.
The directive specifically targets the two models, with Anthropic attributing the government’s action to reported vulnerabilities. According to the company, US officials provided verbal evidence regarding a potential narrow, non-universal jailbreak method for Fable 5, shared by an unnamed entity. While the government did not specify the exact nature of the security risk, Anthropic stated it would share further details on the concerns within 24 hours.
Fable 5 was launched on 9 June to bring capabilities from Anthropic’s state-of-the-art cybersecurity model, Mythos 5, to the public. Mythos 5 has previously been available only to Project Glasswing partners. Anthropic noted that Fable 5’s capabilities exceed those of any previous model it has released, citing internal tests where the model beat Pokémon FireRed, whereas the Claude chatbot failed to beat the original Pokémon Red.
In its announcement, Anthropic expressed disagreement with the decision to block a commercial model based on these concerns. The company argued that the action lacks transparency and technical grounding, describing the recall as disproportionate given the narrow nature of the identified vulnerability. Anthropic reiterated its position that regulatory processes for blocking AI models should be transparent, fair, clear, and grounded in technical facts.
The company highlighted its defence strategy, which aims to make jailbreaks either narrow or very expensive to produce, combined with thorough monitoring to detect and shut down attacks. Anthropic noted that its safeguards for Fable 5 are so strong that many users have complained they are overly broad, and emphasised that perfect resistance to jailbreak attempts is impossible to ensure for any model.
Anthropic confirmed that its other models and the Claude chatbot remain unaffected by the suspension. The company has been vocal about the need for AI safety and regulatory frameworks, warning in the past that applying strict standards across the industry could effectively halt new model deployments for frontier providers.
