US allows Anthropic to release Mythos AI to 'trusted' US organizations
The US government has authorized AI company Anthropic to release its new AI model, Mythos, to "trusted" organizations within the United States, according to a Semafor report. The approval reflects ongoing US efforts to balance AI advancement with national security concerns by controlling access to powerful AI systems.
Background
- Anthropic is an American AI safety company (creator of Claude), known for its focus on responsible AI development. "Mythos" appears to be a powerful new AI model from Anthropic.<br>- The US government (likely via agencies like the Commerce Department or National Security Council) is explicitly permitting — and possibly regulating — which organizations can access Mythos, restricting it to "trusted" US entities only.<br>- This marks an escalation in AI export controls: governments are now controlling *access within their own borders*, not just blocking foreign adversaries. It signals that AI models are being treated like sensitive military technology (similar to nuclear or cryptographic controls).<br>- The story (from Semafor, cited by Reuters) suggests the government assessed Mythos as posing risks serious enough to warrant domestic licensing — a first for a major frontier AI model.