Anthropic's Fable 5 is back worldwide after a two-week government ban over a jailbreak

The 99-plus percent block rate on the jailbreak technique sounds reassuring, but the acknowledged rise in false positives on benign requests is a real operational cost that will affect enterprise customers who built workflows on Fable 5 during its prior availability window.

This story is the technical backstory to what TechCrunch and The Verge both covered on the same date: the Trump administration lifting restrictions and the Department of Commerce greenlighting Fable 5's return. Those pieces framed the reinstatement as a policy negotiation win for Anthropic. This Decoder report reframes it: the ban was not purely geopolitical friction but was also triggered by a concrete safety failure, which means Anthropic had to satisfy both a regulatory and a technical bar simultaneously. That dual-track pressure is worth tracking as a precedent, because it suggests future model suspensions could be faster and harder to negotiate around if the underlying exploit is systemic across a model family rather than isolated.

Watch whether enterprise customers report measurable increases in false-positive rejections over the next four to six weeks. If complaints surface publicly, it will confirm that Anthropic accepted a real capability regression to clear the regulatory hurdle, not just a theoretical one.