More than 50 cybersecurity professionals went public against the US government's ban on Anthropic's two most capable AI models, setting up an industry-versus-policy confrontation that will shape how frontier AI tools are regulated for enterprise security teams globally.
- 54 cybersecurity professionals signed an open letter demanding the US government lift its export control ban on Anthropic’s Fable 5 and Mythos 5 models, calling the directive a threat to national cyber defense.
- The letter targets Commerce Secretary Howard Lutnick and National Cyber Director Sean Cairncross, arguing the ban disarms defenders while adversaries face no equivalent restriction.
- Equivalent capabilities already exist in OpenAI’s GPT-5.5, Claude Opus, Claude Sonnet, and Chinese models including Kimi 2.7, making the ban ineffective against adversaries, the signatories argue.
- Alex Stamos, Joe Levy, and Sandra McLeod are among the signatories, alongside representatives from Nvidia and Adobe.
What Happened
Fifty-four cybersecurity professionals signed an open letter demanding the US government reverse its export control ban on Anthropic’s Fable 5 and Mythos 5 models, calling the directive a direct threat to national cyber defense.
The US Commerce Department issued the ban on June 12, cutting off both models for foreign nationals and forcing Anthropic to disable them for all customers globally. Research from the UK AI Safety Institute, which reported substantial progress toward a universal jailbreak of Fable 5, reportedly triggered the directive.
The open letter, published at freefable.org and addressed to Commerce Secretary Howard Lutnick and National Cyber Director Sean Cairncross, argues the cited capability already exists in OpenAI’s GPT-5.5 and Chinese models like Kimi 2.7. Signatories include Alex Stamos, former chief security officer at Facebook and Yahoo, Joe Levy, CEO of Sophos, and Sandra McLeod, CISO at Zoom Communications, alongside representatives from Nvidia and Adobe.
Why It Matters
Enterprise security teams that had integrated Fable 5 into their workflows lost access overnight with no transition window, a disruption the letter frames as weakening defenders while adversaries face no equivalent restriction. Beyond the operational hit, the signatories warn the action has created market uncertainty and traded away US AI leadership without a proportionate security justification.
The government’s concern is not entirely without basis, given that frontier-level code analysis capabilities present genuine asymmetric risk if made available globally without oversight. What the letter challenges is the method: a blanket shutdown triggered by a narrow, non-universal jailbreak finding, with no prior consultation with the cybersecurity industry that had already flagged Fable 5’s guardrails as too restrictive for legitimate defensive work.
Enterprise customers using Fable 5 or Mythos 5 for production workflows are currently operating without access, with no confirmed timeline for restoration.
Rather than foster resilience, this move creates chaos. William Wright, CEO, Closed Door Security
Bottom Line
The immediate signal to watch is whether Lutnick or Cairncross formally engage with the freefable.org letter, which would open a channel for revising the export control standard. If neither responds, Anthropic’s path likely runs through the courts, a route the company has already used in its dispute with the Pentagon.
For security teams, the policy fight and the operational need run in parallel. Fable 5 is offline, but frontier-level vulnerability scanning is not, and teams waiting on this ban to lift before deploying AI for defensive security work are already ceding ground.
