Cyber Experts Urge White House To Reverse Ban On Anthropic’s Fable Model
The U.S. government ordered Anthropic to block foreign access to its AI model Fable, citing concerns over cyberattack capabilities, but 150 cybersecurity experts urged a reversal, warning it weakens defenses. Critics argue malicious actors already have comparable tools, like OpenAI’s GPT-5.5, and open-weight models will soon surpass Fable’s capabilities.
The U.S. government recently directed Anthropic to restrict access to its Fable AI model, preventing foreigners from using it after Amazon researchers claimed the model’s safeguards could be bypassed for cyberattacks. Fable, a public version of Mythos, was designed with guardrails to limit malicious use, but the government’s order followed reports that attackers could exploit it despite these restrictions. Anthropic complied by shutting down Fable entirely, raising concerns among cybersecurity experts. Over 150 security professionals, including former DHS researcher Jack Cable, signed an open letter opposing the ban, arguing it harms defenders who rely on advanced AI to identify vulnerabilities. Cable, now CEO of cybersecurity startup Corridor, stated that blocking access to the most capable model undermines efforts to protect systems, while adversaries already use similar tools like OpenAI’s GPT-5.5, which is publicly available and cheaper. AI security expert Oege de Moor, founder of Xbow, tested Mythos and found that GPT-5.5 performs nearly as well in bug detection, making the Fable ban ineffective. He warned that open-weight models, such as China’s Kimi-K2.6, will soon match or exceed Mythos’ capabilities, rendering government restrictions obsolete. Once these models are released, he said, no order can reverse their availability. Qasim Mithani, CEO of Depthfirst, a $580 million cybersecurity startup, emphasized that relying on a single AI model is risky due to policy shifts or access revocations. His company develops its own hacking models, claiming they rival or surpass Mythos in some areas. Mithani urged organizations to prioritize autonomy and control over dependence on restricted AI tools. Neither Anthropic nor the U.S. government has updated the public on discussions following the ban. Anthropic initially suggested the order stemmed from a misunderstanding. The cybersecurity community continues to push for reconsideration, arguing that the move benefits adversaries while weakening defensive capabilities.
This content was automatically generated and/or translated by AI. It may contain inaccuracies. Please refer to the original sources for verification.