Anthropic Restores Claude Fable 5 After U.S. Lifts Jailbreak-Linked Export Controls

Key Points:

• Anthropic is restoring worldwide access to its AI model Claude Fable 5 after the U.S. Commerce Department lifted export controls imposed less than three weeks earlier due to security concerns.
• The initial shutdown was triggered by a jailbreak discovered by Amazon researchers that allowed the model to bypass safety rules and demonstrate software vulnerabilities, leading to emergency government restrictions.
• Anthropic addressed the issue by deploying a new safety filter that blocks the reported jailbreak technique in over 99% of cases, rerouting flagged requests to a less capable model and informing users, though this causes some false alarms.
• Access to Mythos 5, a similar model with fewer safety guardrails, remains limited to select U.S. companies and federal agencies, with ongoing government negotiations to expand availability.
• The episode highlights ongoing tensions between AI innovation, security risks, and regulatory frameworks, with Anthropic proposing a new system to classify jailbreak severity and collaborating closely with the government on future model safety and deployment.