A collective of dozens of prominent cybersecurity experts, including several renowned industry veterans, has issued an open letter to the U.S. government, urging the immediate lifting of an export control order placed on Anthropic’s advanced Fable and Mythos models.
The open letter contends that this governmental action has "taken the best models away from [cybersecurity] defenders," thereby preventing them from utilizing these powerful tools to identify vulnerabilities and enhance the security of their software and products.
The letter further emphasizes the critical risk, stating, “To pull the best capabilities away from defenders without a good reason when our adversaries are rapidly advancing is dangerous.”
The U.S. government imposed the export limitation on Anthropic’s Fable and Mythos last Friday, citing unspecified national security concerns, an explanation Anthropic noted lacked specific details. In response to this directive, the company promptly suspended global access to both models for all users.
At the time of its publication, the letter had garnered 76 signatures from cybersecurity specialists. Notable signatories include Alex Stamos, former chief of security at Facebook; Casey Ellis, founder of the bug bounty platform Bugcrowd; Jon Callas, a celebrated cryptographer and former Apple security design and architecture manager; computer scientist Paul Vixie; Dino Dai Zovi, former head of applied security engineering at Block; Katie Moussouris, founder of Luta Security; and Rachel Tobac, CEO of security awareness training firm SocialProof Security.
When Mythos was initially previewed in April, Anthropic asserted its exceptional power in detecting security vulnerabilities. The company consequently implemented stringent access restrictions to mitigate the risk of malicious actors or foreign adversaries exploiting it to cause widespread disruption online. This meant initial access to Mythos was granted to approximately 50 companies, a group later expanded to around 150 organizations across 15 countries.
Anthropic subsequently released Fable last week, positioning it as a public iteration of Mythos equipped with strict guardrails. These measures were designed to prevent its application in biology, chemistry, and cybersecurity, and to hinder the model's distillation for recreation. However, these guardrails proved so restrictive that many cybersecurity experts reported Fable effectively blocked almost any cybersecurity-related prompts.
Anthropic has suggested that the White House's export control order might have been prompted by a report alleging a method to bypass, or "jailbreak," Fable, thereby unlocking its full Mythos-level capabilities.
Katie Moussouris, one of the signatories of the open letter, confirmed that such a method was indeed demonstrated by Amazon researchers in an unpublished paper which she has reviewed.
However, Moussouris countered in a blog post that the paper did not, in fact, present a genuine jailbreak. Instead, she clarified that researchers simply instructed Fable to rectify open-source code containing both publicly known and "deliberately planted vulnerabilities," after the model initially declined to "review the code for security issues."
Moussouris articulated, “The behavior described in the paper cannot meaningfully be fixed, and any attempt would only weaken the model for defense.” She further argued, “Defenders need to be able to ask AI to fix the bugs in a file, explain why the fix matters, and write tests that confirm the patch works. That is not a guardrail bypass. It is the most valuable thing an AI model can do for defensive security: executing the find, fix, and test loop defenders run every day.”
Moussouris's critique found resonance in the open letter, where the expert group collectively stated their belief that the method described in the Amazon paper "can be replicated" across other prominent AI models, including OpenAI’s GPT-5.5, Anthropic’s publicly available Claude Opus 4.8 and Sonnet, and even Chinese models like Kimi 2.7.
Additionally, the letter advocates for the implementation of transparent and equitably enforced regulations, established through "a democratic rule-making process." These regulations, it proposes, should be grounded in scientific research conducted by industry and academic experts, and applied "only to the minimal extent necessary to ensure the safety of the American public."
The Editorial Staff at AIChief is a team of professional content writers with extensive experience in AI and marketing. Founded in 2025, AIChief has quickly grown into the largest free AI resource hub in the industry.