Key facts
- The U.S. government has lifted export controls on Anthropic's Claude Mythos 5 AI model.
- The decision allows Anthropic to release the model to over 100 U.S. institutions, including major companies and government agencies.
- The export controls were initially imposed due to concerns about the model being 'jailbroken' for malicious purposes.
- Commerce Secretary Howard Lutnick stated that appropriate safeguards are now in place.
- Anthropic has committed to working with the U.S. government on protocols and standards for its models.
The U.S. government has lifted its block on Anthropic's Claude Mythos 5 AI model, allowing the company to release it to more than 100 U.S. institutions, including major companies and government agencies, according to a Semafor report. The decision, communicated in a letter from Commerce Secretary Howard Lutnick to Anthropic's chief compute officer Tom Brown, signifies a de-escalation following the administration's imposition of export controls two weeks prior.
The initial restrictions were prompted by warnings that the model, and its cousin Fable 5, could be 'jailbroken' for malicious purposes. Reports suggest Amazon alerted the government to a jailbreak concerning Fable 5, with an adviser to President Trump alleging that Anthropic's CEO Dario Amodei initially dismissed the risk. These concerns led the administration to direct Anthropic to limit access to U.S. citizens, a move that prompted Anthropic to withdraw the models from the market entirely.
Secretary Lutnick stated that appropriate safeguards are now in place and cited "significant progress" in daily talks between the government and Anthropic. The company has committed to collaborating with the U.S. government on protocols and standards for its models. The new arrangement means a license will no longer be required for certain trusted partners to access Claude Mythos 5, including foreign national employees of identified entities.
This development occurs on the same day OpenAI released its GPT-5.6 model to a select group of government-approved partners. The previous restrictions on Mythos were partly driven by suspicions of China-linked groups accessing the model, posing potential national security risks and opportunities for reverse-engineering.