Anthropic announced that the U.S. government has lifted export restrictions on its most advanced AI models, Claude Fable 5 and Mythos 5. This decision reverses an abrupt ban imposed on June 12, just days after the models' release, due to national security concerns related to potential "jailbreaking" or bypassing of safety protocols. The Commerce Department, led by Secretary Howard Lutnick, confirmed the lifting of the ban in a letter to Anthropic, citing the company's agreement to proactively detect and address security risks, collaborate on future AI releases, and report malicious activity.
Fable 5, a consumer-focused AI, and Mythos 5, tailored for businesses and cybersecurity experts, were both suspended after Amazon researchers identified a method to bypass Fable 5's safeguards, allowing it to generate exploit code. Anthropic disagreed with the initial ban, arguing that a narrow jailbreak shouldn't halt a commercial model used by hundreds of millions. In response, Anthropic has trained an improved safety classifier that blocks the specific jailbreak technique in over 99% of cases, rerouting blocked requests to an older model, Opus 4.8.
Anthropic plans to restore global access to Fable 5 on July 1 across its platforms, while Mythos 5 will be reintroduced to a pre-approved list of U.S. organizations. The company has also partnered with Amazon, Microsoft, and Google to develop a framework for assessing AI jailbreak severity, established a 24/7 monitoring team for jailbreak submissions, and will launch a HackerOne program. These measures are part of a broader collaboration with the U.S. government, including providing early access to models for evaluation and sharing threat intelligence, marking a shift towards tighter government oversight on frontier AI models.
