Anthropic announced on Friday that the US government had authorized a limited number of American cybersecurity companies to use its advanced AI system Mythos 5. Earlier restrictions on the model stemmed from national security concerns. The company stated it will keep engaging with officials to broaden availability of Mythos 5 and restore public access to Fable 5. On June 12 authorities required Anthropic to suspend access to two leading AI models after identifying weaknesses in misuse-prevention measures. The decision prompted claims of excessive government intervention. Commerce Secretary Howard Lutnick noted in correspondence that Anthropic had cooperated with federal efforts to mitigate risks tied to the models and that meaningful advances had resulted. A department representative added that work continues to maintain US leadership in AI while protecting security interests. Anthropic had previously declined to supply its technology for mass surveillance or autonomous weapons programs, resulting in canceled Pentagon agreements. Competitor OpenAI introduced GPT-5.6 on the same day under government-approved client restrictions. OpenAI chief Sam Altman remarked that the current approach is not ideal yet expressed general alignment with shared objectives. Earlier this month President Trump issued an executive order establishing voluntary federal assessments of national security risks for advanced AI systems prior to release. Details on enforcement and applicable models remain limited.
