Anthropic Disables Mythos 5 and Fable 5 Under US Pressure

⚡

Key Takeaways

1Anthropic suspended Mythos 5 and Fable 5 three days after their launch, following a U.S. national security directive.

2The Trump administration demands that these AIs be inaccessible to foreigners, even employees of Anthropic, a constraint deemed unenforceable by the startup.

3The decision, initiated by Secretary of Commerce Howard Lutnick, is based on fears of users circumventing safeguards.

💡Why it matters — This suspension illustrates the tensions between technological innovation and government regulation, potentially impacting the entire industry.

Barely launched, the artificial intelligence models Claude Fable 5 and Claude Mythos 5 from Anthropic are already on hold. The American startup announced on June 12 the suspension of these systems to comply with a directive from the U.S. government, citing national security reasons.

The Trump administration required Anthropic to block access to these models for all foreign nationals, including those present in the United States and even non-American employees of the company. Anthropic stated that it is currently impossible to distinguish users by their nationality, which led the company to temporarily disable both models.

The directive, received at 5:21 PM local time, was not accompanied by detailed explanations regarding the security risks. According to Axios, Secretary of Commerce Howard Lutnick made this decision after learning that a company had circumvented safeguards of Mythos 5 and Fable 5, allowing for potentially malicious uses.

Anthropic disputes this justification, claiming that the identified vulnerabilities are less concerning than those of some competitors like OpenAI. The startup even mentions a possible “misunderstanding” with U.S. authorities. It fears that if these requirements were to apply to the entire industry, it could hinder the deployment of future advanced models. It is actively working to restore access to its services.

Claude Fable 5, launched on June 9, is the first public version of Anthropic's Mythos range. Although it is restricted to limit certain sensitive uses, particularly in cybersecurity, biological and chemical weapons, it was already redirecting some queries to Claude Opus 4.8. Claude Mythos 5, less restricted, was reserved for authorized enterprises and performed well in identifying cybersecurity vulnerabilities.

The decision to restrict access to Claude Mythos 5 to a circle of trusted partners had already drawn criticism. Some observers accused Anthropic of engaging in “fear marketing” by highlighting the potential risks of its systems.

The suspension has sparked international reactions, with concerns expressed by countries such as India, the United Kingdom, and China.

Anthropic Disables Mythos 5 and Fable 5 Under US Pressure

Le brief IA que les pros lisent chaque soir

Brief IA — L'actualité IA en français