Claude Fable 5 and Mythos 5 restored globally after US lifts 19-day export ban
The US Commerce Department lifted the export controls it imposed on June 12 after Amazon researchers found a security jailbreak; Anthropic redeployed Fable 5 on July 1 with a new cybersecurity classifier that blocks the triggering prompt
Add to a list
No lists yet.
developments
-
Early enterprise reports flagged the new cybersecurity classifier creating false positives on legitimate developer tasks: GitHub issue #67441 documents failures on SSH server administration prompts and multi-step PDF processing tasks that the classifier incorrectly flags as exploit-adjacent. Anthropic acknowledged the issue in a product notice and said a calibration update would follow before the full-limit restoration on July 7. Separately, during the 19-day suspension window, GPT-5.5 scored above Fable 5 on the ALE (Agent Lifecycle Evaluation) benchmark, a benchmark that Fable 5 had led at launch; the gap narrowed to 2.1 points on the overall index. The episode is cited by enterprise AI buyers as evidence that invisible model switches in agentic pipelines create compliance exposure: contracts that specified Fable 5 ran on a different model for 19 days without customer notification. ↗
Summary
Anthropic redeployed Claude Fable 5 and Mythos 5 globally on July 1, 2026, after the US Commerce Department lifted a June 12 export-control order that had taken both models offline worldwide. The trigger for the ban was a jailbreak discovered by Amazon researchers: a prompt that could bypass safety guardrails and direct the model to identify software vulnerabilities and write exploit code. Anthropic's fix is a new cybersecurity classifier running at inference time, developed with intelligence-community partners and reviewed before Commerce reversed the order. Fable 5 returns at up to 50% of weekly usage limits through July 7 for Pro, Max, Team and selected enterprise users; Mythos 5 is restored only for enterprise-verified accounts. Claude Code and Claude Cowork are also back. The 19-day outage is the first documented case of a US government export-control order being used to remove a commercially deployed frontier AI model from global markets.
Why it matters
The episode establishes that the US can take any AI model offline globally with a single administrative order, regardless of whether it has been commercially released. For Anthropic and for the broader industry, it is the first stress-test of the post-export-control regime that took effect after the Trump administration's AI nationalisation policy. The precedent matters for non-US cloud providers, enterprise buyers locked into AI contracts, and every government contemplating its own AI access rules. The cybersecurity classifier fix also sets a template for how safety incidents trigger regulatory action: faster than any court process and without public disclosure of the specific exploit.
What to watch
- Whether the 50% usage-limit ramp through July 7 extends or Fable 5 is fully restored.
- Whether Congress follows through on demanding written justification from Commerce, which would set a disclosure precedent for future orders.
- How the episode influences other governments considering their own AI export or access controls.