After a prolonged negotiation process with the Trump administration, Anthropic's Mythos 5 model has been approved for limited access to a select group of organizations, including cyber defenders and infrastructure providers. This decision comes as a result of Anthropic's efforts to address risks associated with the model, as stated in a letter from Commerce Secretary Howard Lutnick to Anthropic co-founder Tom Brown. The US government has essentially made an exception for Mythos 5, allowing approved organizations to access the model, while maintaining the export control directive that bars foreign nationals from accessing the model. Anthropic employees who are not US nationals and members of approved organizations who are not US nationals are also greenlit to access Mythos 5. The company is working to provision the approved set of providers and restore their access to Mythos 5 as quickly as possible. This development mirrors the deal secured by OpenAI for its GPT-5.6 model, where the model is released in a limited preview setting for approved organizations only. Both AI labs hope for more general availability in the future, but this will ultimately depend on the Trump administration's decisions. The current consensus among tech leaders is that the past two weeks have not been an ideal approach to AI regulation, and a more sustainable framework is needed to ensure the responsible development and deployment of AI models.

Source: The Verge