According to Detections Beating, a monitoring group, over 40 cybersecurity experts, led by former Facebook Chief Security Officer Alex Stamos, have jointly signed an open letter protesting the Trump administration's export ban on Anthropic's most powerful AI model. The open letter points out that forcibly shutting down large models like Mythos and Fable 5 will not fundamentally eliminate security risks but will instead deprive defensive personnel who are combating AI-driven automated hacking offensives of a key weapon.
The White House's export restrictions decision stems from a vulnerability report submitted by Amazon security researchers. The White House is concerned that Fable 5 can automatically generate vulnerability proof-of-concept code and, under specific conditions, evolve into a fully autonomous attack chain. However, Stamos argues that the White House's concerns are misguided, as the standard version of Fable 5 cannot scan the entire Linux kernel and pinpoint all security vulnerabilities. The version alleged to have autonomous attack transformation capabilities is actually limited to the unreleased Mythos 5 designated for internal testing only and Mythos Preview.
The open letter emphasizes that automatically generating vulnerability proof-of-concept code is not unique to Fable 5. Through OpenAI's GPT-5.5, Anthropic's Opus and Sonnet, and even China's Kimi 2.7, similar levels of vulnerability analysis and code generation can be achieved. Stamos warns that Chinese open-source models are already closely following Fable 5 in analyzing system flaws, and at a critical juncture where adversaries are heavily stockpiling and exploiting vulnerabilities, the U.S.'s decision to disable its top-tier domestic defense tool is undoubtedly a self-defeating move.
