Disarming the Defenders: 76 Security Veterans Demand Washington Reverse the Fable 5 Ban

The export-control order that took Anthropic’s most powerful models offline now has an organized opposition. On June 15, a group of 76 cybersecurity professionals — CEOs, CISOs, researchers and investors — published an open letter to Commerce Secretary Howard Lutnick and National Cyber Director Sean Cairncross demanding that the government immediately lift the directive that darkened Claude Fable 5 and Mythos 5. According to TechCrunch, the signatories include some of the field’s best-known names: Alex Stamos, the former chief security officer of Facebook; Katie Moussouris of Luta Security; cryptographer Paul Vixie; Bugcrowd founder Casey Ellis; Jon Callas; Dino Dai Zovi; and SocialProof Security chief executive Rachel Tobac.

Their argument is one of asymmetry. “To pull the best capabilities away from defenders without a good reason when our adversaries are rapidly advancing is dangerous,” the letter states, contending that the action “has taken the best models away from defenders” — the very people who use frontier models to hunt for vulnerabilities and harden systems before attackers reach them. The experts also dispute that the flagged jailbreak is unique to Fable 5, noting per Cybersecurity Dive that the same technique can be reproduced on rival systems including OpenAI’s GPT-5.5, Anthropic’s own Claude Opus 4.8, and Chinese models such as Kimi 2.7.

Rather than a unilateral takedown, the signatories want any restriction to come through “a democratic rule-making process” grounded in scientific research, and enforced only “to the minimal extent necessary to ensure the safety of the American public.” They frame the stakes in competitive terms as much as defensive ones: with “China’s advanced AI models only months behind the best American models,” pulling a leading U.S. system from defenders’ hands hands an advantage to exactly the adversaries the order claims to guard against — while seeding the market uncertainty of a model that can vanish overnight.

Anthropic has sharpened its own position in parallel. The company argues that “if this standard was applied across the industry, we believe it would essentially halt all new model deployments” for every frontier provider, since the ability to find software flaws — the capability at issue — is now common to all of them. The directive landed three days after the models launched as Anthropic’s most capable yet; the company took both offline worldwide because it could not police the order’s foreign-national line user-by-user in real time. (See our earlier coverage of the worldwide suspension and of the Amazon-led push that reportedly set it in motion.)

As of publication the models remain dark, with no public restoration timeline and no documented government response to the letter. What the episode exposes is less a single dispute than a missing process: there is no agreed playbook for how Washington should weigh a contested security finding against a commercial model already serving hundreds of millions of users. The cybersecurity community’s letter is, in effect, an attempt to write the first lines of that playbook before the precedent hardens into the default.