level: technical
a group of 76 cybersecurity experts published an open letter asking the us government to reverse its export control order on anthropic's fable and mythos models. the order, issued on friday citing national security concerns, led anthropic to suspend worldwide access. signatories include former facebook security chief alex stamos, bugcrowd founder casey ellis, and cryptographer jon callas. they argue the ban removes critical tools from defenders who use the models to find and fix software vulnerabilities.
the ban may stem from an amazon research paper that reportedly showed a method to bypass fable's guardrails. however, signatory katie moussouris, who reviewed the paper, says it did not demonstrate a real jailbreak. instead, researchers asked fable to fix code with known and planted vulnerabilities after it initially refused. moussouris argues this is normal defensive work, not a bypass, and that fixing this behavior would weaken the model for security tasks. the open letter claims similar methods work on other models like openai's gpt-5.5 and chinese model kimi 2.7.
the letter calls for transparent, science-based regulations created through a democratic process and applied minimally. it stresses that removing advanced models from defenders while adversaries advance is dangerous. the controversy highlights tensions between national security controls and the need for cybersecurity professionals to access powerful ai tools for defense.
why it matters: the ban restricts ai tools that cybersecurity defenders rely on to find and patch vulnerabilities, potentially leaving systems more exposed to attacks.