A coalition of 76 cybersecurity professionals, including former Facebook CSO Alex Stamos and Bugcrowd founder Casey Ellis, has published an open letter urging the U.S. government to reverse an export control order on Anthropic's Fable and Mythos models. The signatories argue the ban strips defenders of their most effective vulnerability detection tools while adversaries advance unimpeded.

The order, issued Friday without detailed public justification, forced Anthropic to suspend global access to both models. Mythos, previewed in April, was initially restricted to 50 organizations due to its potent vulnerability discovery capabilities—later expanded to 150 entities across 15 countries. Fable, a public release with guardrails blocking biology, chemistry, and cybersecurity queries, arrived last week but proved so restrictive that many security queries were outright rejected.

Anthropic suggests the ban stemmed from a non-public Amazon paper claiming a method to jailbreak Fable and unlock Mythos-level capabilities. Signatory Katie Moussouris, founder of Luta Security, reviewed the paper and disputes the finding. In her analysis, researchers simply asked Fable to patch known vulnerabilities in open-source code—a standard defensive task—after the model refused a direct security review. "That is not a guardrail bypass. It is the most valuable thing an AI model can do for defensive security," she wrote.

The open letter argues the same technique works on OpenAI's GPT-5.5, Anthropic's Claude Opus 4.8, and even China's Kimi 2.7, suggesting the ban is both ineffective and counterproductive. The experts call for transparent, science-based regulations developed through democratic processes, applied minimally to protect national security without crippling defensive capabilities.

Source: TechCrunch