Key facts
- Anthropic disabled its Fable 5 and Mythos 5 AI models globally following a US export control order.
- The US Commerce Department cited urgent security concerns over the models' potential misuse.
- The order specifically banned foreign nationals from accessing the advanced AI models.
- Anthropic stated the US government misunderstood the risks and that similar flaws exist in competitor models.
- The decision has drawn criticism from international allies and raised investor concerns about political risk in the AI sector.
Anthropic has been compelled to disable its most advanced AI models, Fable 5 and Mythos 5, for all users globally following an unprecedented export control order from the U.S. Commerce Department. The order, issued just days after the models' launch, cited urgent national security concerns related to potential misuse by foreign nationals.
Due to the difficulty in restricting access by nationality alone, Anthropic opted to disable the models entirely to ensure compliance. This decision came after reports that the UK's AI Safety Institute had made significant progress in jailbreaking Fable 5, and that Amazon researchers had alerted the U.S. government to the model's potential for cyberattack-related information extraction. Anthropic has contested the severity of the risks, stating that similar flaws exist in rival models and that the U.S. government's actions are overly heavy-handed and could stifle AI development.
The U.S. government's intervention has sparked criticism from international allies, with European Commission spokesperson Thomas Regnier stating that contingency measures should not be discriminatory. British MP Tom Tugendhat also voiced concerns about the impact on British hospitals and companies that had access to the model for vital research. The situation has led to broader discussions about political risk in the AI sector, with some investors viewing Anthropic as a riskier bet due to potential government interference.