A chorus of influential voices from the cybersecurity world is raising serious alarms over a recent U.S. government decision. Dozens of seasoned experts, including many industry veterans, have penned an open letter urging the lifting of an export control order on Anthropic’s most powerful AI models: Fable and Mythos. This controversial ban, they argue, is actively undermining global cybersecurity efforts.
According to the impassioned letter, this government action has stripped vital tools from cybersecurity defenders. They lament that these cutting-edge models, crucial for identifying vulnerabilities and fortifying software, are now inaccessible. The experts warn that “to pull the best capabilities away from defenders without a good reason when our adversaries are rapidly advancing is dangerous.”
The Models Under Scrutiny: Fable and Mythos
The U.S. government issued its export control order last Friday, directing Anthropic to restrict the export of Fable and Mythos. Citing undisclosed national security concerns, the government offered no specific reasons for the ban, prompting Anthropic to suspend global access to these advanced models.
Anthropic’s Mythos, first previewed in April, quickly gained a reputation for its extraordinary ability to pinpoint security vulnerabilities. Its power was so immense that the company initially restricted access to prevent misuse by malicious hackers or foreign adversaries, granting it to just 50 companies, later expanding to 150 organizations across 15 countries.
Last week saw the release of Fable, a public iteration of Mythos, equipped with stringent guardrails. These safeguards were designed to prevent its application in sensitive areas like biology, chemistry, and cybersecurity, as well as to deter attempts at model distillation. However, many cybersecurity experts found Fable’s guardrails to be excessively strict, blocking virtually any prompt related to cybersecurity tasks.
Cybersecurity Veterans Sound the Alarm
The open letter, signed by an impressive 76 cybersecurity experts, features some of the biggest names in the field. These include Alex Stamos, former chief of security at Facebook; Casey Ellis, founder of the bug bounty platform Bugcrowd; and Jon Callas, famed cryptographer and former Apple security design manager. Other prominent signatories are 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 SocialProof Security).
Their collective concern highlights a perceived misstep by the government, which they believe jeopardizes defensive capabilities. The experts emphasize the paradox of hamstringing defenders while sophisticated threats continue to evolve at an alarming pace.
Debunking the “Jailbreak” Claim
The White House’s export control order may have stemmed from a report alleging a method to bypass—or “jailbreak”—Fable’s guardrails, thereby unlocking its full Mythos-level capabilities. Katie Moussouris, one of the letter’s signatories, reviewed an unpublished Amazon paper reportedly demonstrating this method, but her findings contradict the “jailbreak” narrative.
Moussouris contends that the paper did not showcase a genuine jailbreak. Instead, researchers simply instructed Fable to rectify known and deliberately planted vulnerabilities within open-source code, after the model initially declined a general security review. She argues that this isn’t a guardrail bypass, but rather a fundamental and essential function for defensive security operations.
Moussouris elaborated that “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.” She further asserted that such defensive capabilities cannot be meaningfully ‘fixed’ or limited without weakening the model’s utility. Moreover, the experts’ letter points out that these supposed “jailbreak” capabilities can be replicated on other widely available models, including OpenAI’s GPT-5.5, Anthropic’s Claude Opus 4.8 and Sonnet, and even Chinese models like Kimi 2.7.
A Call for Transparent and Scientific Regulation
The open letter doesn’t just critique the ban; it also calls for a more thoughtful approach to AI regulation. The signatories advocate for transparent and fairly enforced regulations that are developed through a democratic rule-making process.
They insist that any new rules must be grounded in robust scientific research from industry and academic experts. Crucially, these regulations should be applied only “to the minimal extent necessary to ensure the safety of the American public.” This collective plea underscores the need for collaboration between policymakers and technical experts to navigate the complexities of advanced AI in a way that truly enhances national security without inadvertently undermining it.
Source: TechCrunch – AI
