Fable 5 Is Back: Anthropic Restores Access After 19-Day Government Ban

Anthropic restored Claude Fable 5 globally on July 1 after the US lifted export controls. A new safety classifier blocks the reported jailbreak.

AI Tutorials · · 2 min read

Quick answer

Anthropic restored global access to Claude Fable 5 on July 1, 2026, after the US Commerce Department lifted export controls that had kept the model offline for 19 days. A new safety classifier now blocks the reported jailbreak in over 99% of cases. Fable 5 is available again across Claude.ai, Claude Code, and the API.

The most dramatic AI shutdown in history is officially over. Anthropic restored global access to Claude Fable 5 on July 1, 2026 — 19 days after the US Commerce Department ordered the model taken offline over a reported jailbreak vulnerability.

Fable 5 and its larger sibling Mythos 5 are now available again across Claude.ai, Claude Code, Cowork, and the API. The restoration follows the Commerce Department’s decision on June 30 to lift the export controls it had imposed on June 12, just three days after the models launched.

What Changed

The key difference is a new safety classifier. During the shutdown, Anthropic trained a smaller AI system that monitors conversations and detects when Fable 5 is asked to perform potentially harmful cybersecurity tasks. The classifier blocks the specific jailbreak — originally discovered by Amazon researchers — in over 99% of cases.

When the classifier triggers, users see a notification and their request is automatically rerouted to Opus 4.8 instead. There is a trade-off: Anthropic acknowledges that the tighter filtering may also block some legitimate coding and security requests.

The Deal With the Government

In exchange for lifting the ban, Anthropic agreed to three conditions: proactively detect and address security risks in its models, work with the government on testing protocols and release standards for current and future models, and report any observed malicious activity.

This framework could set the template for how future AI model disputes between companies and regulators are resolved — a negotiated set of safeguards rather than an outright ban.

What This Means for You

If you were using Fable 5 before the shutdown, you can pick up where you left off. The model retains its position at the top of coding benchmarks, scoring 80% on SWE-Bench Pro. For most users, the experience will be identical — the new classifier runs in the background and only activates on specific cybersecurity-related prompts.

If you are new to Fable 5, now is a good time to try it. It remains one of the most capable models available for complex reasoning, coding, and creative tasks.

The 19-day saga is a reminder that even the most powerful AI tools exist within a regulatory framework — and that framework is still being written. For a full timeline, see our earlier coverage of the original shutdown and Anthropic’s signals that a return was imminent.

Frequently asked questions

Is Claude Fable 5 available again?
Yes. Anthropic restored global access to Fable 5 on July 1, 2026, across Claude.ai, Claude Code, Cowork, and the API. Both Fable 5 and Mythos 5 are back online.
What changed about Fable 5 since the shutdown?
Anthropic added a new safety classifier that blocks the specific jailbreak technique reported by Amazon researchers in over 99% of cases. Some benign coding and security requests may also be filtered as a side effect.
Why was Fable 5 shut down in the first place?
The US Commerce Department issued an export control directive on June 12, 2026, after Amazon researchers reported a jailbreak vulnerability. The government ordered Anthropic to block all foreign national access, which forced a total global shutdown.
What did Anthropic agree to in exchange for lifting the ban?
Anthropic agreed to proactively detect and address security risks, work with the US government on testing protocols and release standards for future models, and report any observed malicious activity.

Want to keep learning?

Explore our guided learning paths or try building something with AI right now.

Enjoyed this article?

Subscribe for more AI insights delivered to your inbox every week.

No spam. Unsubscribe anytime.