US Lifts Export Controls on Anthropic's Fable AI After Two-Week Freeze
The Commerce Department cleared Anthropic's most powerful AI model for international customers following a government review triggered by fears the system could be used to assist serious cyberattacks.
Key points
- Anthropic, a San Francisco AI company, suspended foreign access to its Fable 5 and Mythos 5 models for more than two weeks after US government safety concerns.
- US export controls, rules that restrict the sale of sensitive technology to foreign customers, were placed on both models over fears they could help criminals or foreign states carry out serious cyberattacks.
- Commerce Secretary Howard Lutnick announced on social media that his department had reviewed and approved Fable 5 for release.
- Anthropic restored customer access after the Commerce Department confirmed it had worked directly with the company to verify the model was safe to export.
- No customers were reported to have had their data accessed; the restriction was a sales and access block, not a security breach.
Anthropic, an American company that builds large AI systems, was forced last month to cut off all non-US customers from its two most advanced products: Fable 5 and Mythos 5.
The reason was export controls. The US government restricts who can receive certain powerful technologies because, in the wrong hands, they could cause serious harm. The Commerce Department applied that logic here, concluding that Fable 5 was powerful enough that a foreign government or criminal group might use it to plan sophisticated cyberattacks on critical infrastructure.
The blackout ran just over two weeks.
Why did the US government freeze access to an AI chatbot?
US officials were not worried about ordinary users asking Fable 5 to write emails. They were concerned its advanced reasoning ability could give a well-resourced attacker meaningful help in finding weaknesses in computer systems or planning large-scale intrusions, tasks that normally require highly skilled human experts.
Export control rules, codified under the Export Administration Regulations, give the Commerce Department broad authority to restrict technology that could offer foreign parties a military or intelligence advantage. Applying those rules to a commercial AI model is relatively new territory.
Commerce Secretary Howard Lutnick said in a social media post, quoted by Anthropic and first reported by The Guardian Technology, that his department had spent the two weeks working closely with Anthropic to "analyze and approve Fable 5 to ensure alignment across the US Government and strengthen America's leadership in AI." Anthropic cited that post when it announced the restoration of access.
The Mythos 5 model's status was not separately addressed in Lutnick's statement. Anthropic hasn't publicly clarified whether Mythos 5 remains subject to any restrictions, a detail worth watching given that our 13 June report on the original geofencing order noted Anthropic disagreed with the government's assessment but complied anyway.
Should you worry about your AI tools being switched off?
For international businesses or researchers who rely on Anthropic's tools, access is restored. The episode signals, though, that advanced AI models now sit inside the same regulatory framework that governs military hardware exports. Future models may face similar reviews before foreign customers can reach them.
Organisations that depend on third-party AI services for any part of their operations should treat sudden access blackouts as a genuine operational risk. The question isn't whether governments will scrutinise these models again. It's how long the next freeze lasts.
