Anthropic restores access to Claude Fable 5
Anthropic restores Claude Fable 5 access after US export controls lifted See how Anthropic blocked a safeguard bypass and expanded AI security work
Anthropic said it will restore global access to Claude Fable 5 on July 1 after US export controls that briefly restricted use of the model were lifted on June 30. The company also said access to Claude Mythos 5 has been partially restored for some US organizations, with broader reenablement still in progress.
The post explains that the temporary suspension followed a government order tied to concerns about cybersecurity safeguards. Anthropic said it worked with US officials and partners, including Amazon, to review a reported safeguard bypass and to update its safety classifier so the specific technique is blocked in most cases.
Anthropic also used the update to outline its broader approach to model safety, including its use of classifierbased defenses, its view that jailbreaks vary in severity, and a proposed industry framework for assessing jailbreak risks. The company said it is expanding cooperation with US government agencies on prerelease testing, information sharing, and joint AI security research.