Anthropic's safety warnings may have just backfired - the government has pulled the plug on its most powerful AI

Key Points:

• The U.S. government ordered Anthropic to immediately disable access to its two most powerful AI models, Claude Fable 5 and Claude Mythos 5, citing national security concerns, affecting all users worldwide.
• Mythos 5, known for its exceptional ability to detect security vulnerabilities, was only shared with about 50 vetted organizations for defensive cybersecurity, while Fable 5 was a safer, publicly released version with guardrails against high-risk outputs.
• The government’s directive is framed as an export control action but appears linked to a claimed jailbreak of Fable 5, which Anthropic disputes, stating the evidence is verbal and the capability is already available in other public AI models.
• Anthropic argues its independent classifier safeguards prevent dangerous outputs even if the model is prompted beyond refusals, and it believes the government’s action could stifle AI innovation if applied industry-wide.
• The move disrupts Anthropic’s business at a critical time as it plans an IPO and highlights the irony that its cautious approach to AI safety has attracted government scrutiny.