US Government Orders Anthropic to Pull Fable 5 and Mythos 5 Over Security Concerns

The US government issued an export control directive ordering Anthropic to immediately suspend access to its Fable 5 and Mythos 5 AI models. Authorities cited national security concerns linked to a suspected jailbreak capable of bypassing the models' safety guardrails. Anthropic disputes the severity of the threat but complied with the directive and disabled both models for all users globally.

What the Directive Required

Anthropic said it received the directive at 5:21 pm ET on Friday, June 13. The government ordered the company to suspend Fable 5 and Mythos 5 access for any foreign national, regardless of location, including foreign national Anthropic employees.

To ensure full compliance, Anthropic disabled both models for every user worldwide. The shutdown covers only Fable 5 and Mythos 5. All other Anthropic models, including Opus 4.8, continue to operate normally. No restoration timeline was given.

Anthropic's public statement was brief: "We are complying with the government's legal directive and are removing access to Fable 5 and Mythos 5 for all users."

The Suspected Jailbreak

Anthropic said the government gave no written evidence of the threat, only a verbal description of a potential "narrow, non-universal jailbreak." The technique, as described, involves prompting the model to read a specific codebase and identify or fix software flaws.

Anthropic calls the risk overstated. A narrow jailbreak works only in a specific context, unlike a universal one that bypasses safety measures across all inputs. The company said applying this standard across the AI industry "would essentially halt all new model deployments for all frontier model providers." It believes the directive results from a misunderstanding and is working to restore access.

How Amazon Triggered the Crackdown

Reports indicate that a warning from Amazon prompted the government action. The specific nature of Amazon's signal has not been disclosed, but reports describe a chain: Amazon raised a concern, the government issued the directive, Anthropic complied. Fable 5 and Mythos 5 launched just days earlier and represent Anthropic's most capable models to date.

The shutdown affects every product integrating these models, including tools in the crypto space. AI agents for trading, smart contract auditing, and crypto wallet management rely heavily on Claude API. Days before the suspension, Mythos AI completed an independent audit of the Zcash blockchain and found no serious vulnerabilities.

Implications for AI in Crypto

Anthropic does not contest the legality of the directive but disagrees with its basis and is pushing for quick resolution. The company received no written evidence of the threat, only a verbal account.

For crypto developers and protocols that depend on Claude for security reviews or automated agents, even a short-lived suspension creates operational risk. Hardware wallets and offline storage solutions are unaffected by AI-service outages, but most modern DeFi tools now integrate large language models for risk analysis and automation. The Anthropic case shows that even well-resourced AI providers can face sudden service gaps due to regulatory action.