Anthropic restores Claude Fable 5 with new security guardrails

Anthropic has restored customer access to the Claude Fable 5 model via the Claude API and other platforms on July 1, three weeks after suspending it due to a U.S. Department of Commerce export control directive.

• The suspension followed Amazon researchers using the model to identify cybersecurity vulnerabilities, prompting the government to cite national security concerns.
• As part of the agreement with the U.S. government, Anthropic added guardrails that block certain cybersecurity queries and route them to the less capable Claude Opus 4.8.
• Access for AWS, Google Cloud, and Microsoft Foundry customers was restored shortly after the global availability announcement.

The redeployment resolves a high-profile conflict between the U.S. government and AI providers regarding security assessments of powerful models.