On July 1, Anthropic brought Claude Fable 5 back to the whole world. It’s not just another launch: it’s the end of an unprecedented episode in the industry — 19 days during which the company’s most powerful model was switched off by order of the US government. Here’s what happened, what changed and what it means for you, with verified sources.
What happened: a 19-day blackout
- On June 12, the US government applied export controls to Claude Fable 5 and Claude Mythos 5, requiring access restrictions for foreign nationals. Since Anthropic had no reliable way to verify nationality in real time, it suspended access to both models for everyone, according to Anthropic’s official announcement.
- The trigger: a report by Amazon researchers who found a way to bypass Fable 5’s safeguards to make it identify software vulnerabilities — a sensitive cybersecurity capability.
- On June 30, the US lifted the controls, as confirmed by CNBC, and on July 1 Fable 5 became available globally again.
What changed in the model
- New cybersecurity classifier: Anthropic trained, in collaboration with the government, an improved safety classifier that specifically blocks the type of abuse described in the report.
- Where it’s available: Fable 5 is back on Claude.ai, the API, Claude Code and Claude Cowork for users worldwide.
- Mythos 5 remains restricted: the version with fewer safeguards has only been restored for a small group of approved US organizations (the Glasswing program, defensive cybersecurity), with plans to expand gradually.
The terms of the deal
According to statements by the Commerce Secretary reported in the press, Anthropic committed to proactively detecting and addressing security risks in its models, working with the government on standards for upcoming releases, and reporting malicious activity. In exchange, it won’t need an export license.
Our take
- It’s a huge precedent: for the first time, a government has de facto switched off an already-deployed frontier model, and the industry saw it can happen. “AI as infrastructure” now includes regulatory risk.
- For everyday users, the outcome is good: the model returns with extra safeguards and no announced price changes. If you use Claude, you probably noticed no technical difference on its return.
- The lesson for companies: if your business depends on a single model, this episode is the warning — have a plan B. Diversifying providers stopped being paranoia.
Frequently asked questions
Can I use Fable 5 now?
Yes — since July 1 it’s globally available on Claude.ai, the API, Claude Code and Claude Cowork.
What is Mythos 5 and why is it still limited?
It’s the variant of the same model with fewer safeguards, intended for defensive cybersecurity partners. Only organizations approved by the US government can use it for now.
Conclusion
Fable 5 returns more hardened and with a clear message: the most capable models will live under regulatory scrutiny. For Claude family context, see what Claude Fable 5 is and the recent Claude Sonnet 5 launch; and if the dark side interests you, our guide to AI and cybersecurity.
