The US has lifted export restrictions on Anthropic's advanced AI models, Fable 5 and Mythos 5, three weeks after they were flagged as national security risks. Anthropic confirmed Fable 5 is now available globally, while US organizations regained access to Mythos 5 on June 26. The company is working with the government to expand Mythos access for cybersecurity research under the Glasswing program.

This move follows intense collaboration between Anthropic and the US government to address security concerns. The Commerce Department had initially blocked exports due to fears of misuse by adversaries for cyberattacks. Anthropic responded by enhancing safety protocols, including red-teaming exercises with hackers and a 24/7 monitoring team for jailbreak threats.

While Fable 5 now boasts strengthened safeguards, Anthropic acknowledges a trade-off: some routine coding and debugging prompts may be blocked. The company is developing a framework with industry partners to better assess and respond to AI jailbreak risks, prioritizing efficiency and safety in AI deployment.