76 Security Experts Slam U.S. Export Ban on Anthropic's Top AI Models as a Gift to Adversaries

A coalition of 76 cybersecurity veterans, including former Facebook chief security officer Alex Stamos and Bugcrowd founder Casey Ellis, has published an open letter demanding the U.S. government reverse an export control order against Anthropic's Fable and Mythos models. The group argues the ban strips defenders of their most effective tools at a time when adversaries are accelerating their own capabilities.

"To pull the best capabilities away from defenders without a good reason when our adversaries are rapidly advancing is dangerous," the letter states. The order, issued Friday without detailed public explanation, forced Anthropic to suspend global access to both models. The company had already tightly restricted Mythos since its April preview, limiting it to about 150 organizations across 15 countries due to its power in finding security vulnerabilities. Fable, a public version released last week, was designed with strict guardrails - so strict that many experts found it blocked nearly all cybersecurity prompts.

The White House action may have been triggered by a non-public Amazon paper claiming a method to jailbreak Fable. Signatory Katie Moussouris, founder of Luta Security, reviewed the paper and called the claim misleading. In a blog post, she explained that researchers simply asked Fable to fix known vulnerabilities in open source code - exactly the task defenders need. "That is not a guardrail bypass. It is the most valuable thing an AI model can do for defensive security," she wrote.

The letter also notes the same technique can be replicated on OpenAI's GPT-5.5, Anthropic's Claude Opus 4.8, and even China's Kimi 2.7, questioning the singling out of Anthropic. The group calls for transparent, science-based regulations developed through democratic processes, applied only as necessary to protect public safety.

Source: TechCrunch