Anthropic Restores Global Access to Claude Fable 5 After Export Controls Lifted

Anthropic says access to Claude Fable 5 and Mythos 5 has been restored after US export controls that temporarily forced the company to suspend both models for all users.

Fable 5 will return on July 1 across Claude.ai, the Claude Platform, Claude Code, and Claude Cowork. Pro, Max, Team, and some Enterprise customers will receive access for up to half of their weekly usage limits through July 7; continued use will then rely on usage credits. Anthropic says it is also working to restore availability on AWS, Google Cloud, and Microsoft Foundry.

What changed

The restrictions followed a report from Amazon researchers describing a prompt technique that could make Fable 5 identify software vulnerabilities and generate an exploit demonstration. Anthropic’s follow-up testing found that less capable models from several vendors could perform similar tasks, and said the incident did not reveal unique offensive capabilities associated with Mythos 5.

Still, Anthropic worked with government and industry partners on a new safety classifier. The company says it blocks the reported technique in more than 99% of cases. Requests that trigger the system will be redirected to Opus 4.8, although the stricter filter may also reject more legitimate coding, debugging, and defensive-security work.

Mythos 5 access has been restored for selected US organizations, while Anthropic continues seeking approval to expand availability through Project Glasswing. The company is also collaborating with Amazon, Microsoft, Google, and other partners on a shared framework for evaluating jailbreak severity, alongside deeper government cooperation on pre-release testing and information sharing.

For AI builders, the episode highlights a practical tradeoff: stronger safeguards can reduce misuse but increase friction for legitimate technical users, while regulatory decisions can abruptly change model availability across products and cloud platforms.

Source: Anthropic News