- U.S. officials initially blocked Anthropic's advanced AI models, Fable and Mythos, citing national security risks due to their potential to identify critical infrastructure vulnerabilities.
- An export ban was imposed, which Anthropic found difficult to enforce, leading to negotiations with the US Commerce Department.
- The ban has since been lifted, with Anthropic implementing enhanced safeguards and collaborating with the US government to make its models available.
- Mythos 5 is now accessible to "trusted" US organizations, while Fable 5, designed for the general public with stronger protections, is set for wider release.
- Anthropic acknowledges the challenge of fully preventing AI "jailbreaks" and continues to work with government and industry partners to develop shared safety standards.
IN FULL
Anthropic Mythos: AI system so ‘dangerous’ it was banned is returning
