Cybersecurity experts protest US ban on Anthropic's Fable and Mythos AI models

A coalition of 76 cybersecurity veterans, including former Facebook Chief Security Officer Alex Stamos, Bugcrowd founder Casey Ellis, and cryptographer Jon Callas, has publicly condemned the U.S. government's export control order on Anthropic's most advanced AI models, Fable and Mythos. The open letter warns that removing these powerful tools from defenders while adversaries advance is 'dangerous.' The ban, issued last Friday without detailed explanation, forced Anthropic to suspend global access to the models.

Mythos, launched in preview in April, was touted as exceptionally capable at identifying security vulnerabilities—so much so that Anthropic restricted it to only 150 organizations across 15 countries. The public version, Fable, included strict guardrails blocking use in cybersecurity, biology, and chemistry. Critics argue the claimed jailbreak—demonstrated by Amazon researchers in a non-public paper—is actually a standard defensive task: asking the model to fix known bugs. Katie Moussouris noted that other models like OpenAI's GPT-5.5 and Claude Opus 4.8 can replicate the behavior without needing a bypass. The letter demands transparent, science-based regulation developed through democratic processes.