Summary & Key Takeaways ​

• The US government issued a directive to Anthropic to suspend access to Fable 5 and Mythos 5.
• Access is to be suspended for all foreign nationals, both inside and outside the US.
• The directive cites national security concerns regarding a potential "jailbreak" method.
• Anthropic reviewed the alleged jailbreak and found it identifies minor, known vulnerabilities.
• They state similar capabilities are widely available in other public models, including GPT-5.5.
• The government provided only verbal evidence of the narrow, non-universal jailbreak.
• Simon Willison confirmed his access to Fable 5 was cut off shortly after the directive.

Our Commentary ​

This is genuinely unsettling. A government directive to suspend access to specific AI models, citing national security, feels like a new frontier. We're seeing the real-world implications of AI's power and the scramble for control. Anthropic's counter-argument that the 'jailbreak' is minor and widely available in other models is a critical detail. It makes me wonder if this is truly about security or something else entirely. The speed at which access was cut is also jarring. This is a big one, folks.