The US government issued an export control directive on June 12 requiring Anthropic to suspend access to its flagship AI models, Fable 5 and Mythos 5, for all foreign nationals, regardless of their location. As a result, Anthropic disabled these models globally for non-US users, including its own foreign-national employees, while access to other AI models remains unaffected, according to medianama.com.
Anthropic said the directive was linked to concerns about a possible method to bypass or "jailbreak" Fable 5's safeguards. However, the company stated it was not provided with specific details of the national security concern and only received verbal evidence of a narrow, non-universal jailbreak. Anthropic described the vulnerabilities as minor and previously known, adding that other publicly available models can identify them without requiring a bypass.
The company defended Fable 5's safety measures, claiming its safeguards are substantially more effective than those of any earlier deployed model. Anthropic emphasized that no tester has found a universal jailbreak and acknowledged that perfect jailbreak resistance is not currently achievable. The move highlights ongoing tensions between AI innovation and national security controls in the US technology sector.
Anthropic's suspension of Fable 5 and Mythos 5 access for foreign nationals took effect immediately following the June 12 directive, impacting global users outside the United States and its foreign-national employees, as reported by medianama.com.