Claude Fable 5 came back online globally on July 1, 2026, closing a 19-day blackout that began on June 12 when the U.S. Department of Commerce cited national security authorities to pull the model's export license. Commerce Secretary Howard Lutnick withdrew the requirement on June 30 in a letter viewed by CNBC, and Anthropic re-enabled Fable 5 across Claude.ai, Claude Code, Claude Cowork, and the Claude Platform, with AWS, Google Cloud, and Microsoft Foundry availability following. Pro, Max, Team, and select Enterprise plans get Fable 5 for up to 50% of weekly usage limits through July 7.

The precipitating event was a jailbreak disclosed by Amazon researchers, who prompted Fable 5 into identifying software vulnerabilities and, in one case, producing exploit code. Anthropic's post-mortem is unsentimental about what that actually demonstrated: Claude Opus 4.8, GPT-5.5, and Kimi K2.7 identified the same vulnerabilities, and every model tested could reproduce the exploit demo. The behavior, Anthropic writes, "did not expose any unique Mythos-level cyber capabilities" and "only involved routine defensive cybersecurity work."

That framing matters, because the regulatory response ran ahead of the technical severity. Lutnick had already restored Mythos 5 (same underlying model, fewer safeguards) to a set of U.S. organizations on June 26 after determining "appropriate safeguards" were in place for trusted partners. Fable 5 followed four days later.

Anthropic isn't returning empty-handed. A new safety classifier, trained in coordination with the government, now blocks the reported technique in over 99% of cases; flagged requests are re-routed to Opus 4.8 and users are notified. Anthropic concedes the classifier flags more benign coding and debugging work than its predecessor. Researchers at Commerce's Center for AI Standards and Innovation tested both the prior and new safeguards and, per Anthropic, called them "extraordinarily strong."

The more consequential move is procedural. Anthropic published an early draft of an industry jailbreak-severity framework co-developed with Amazon, Microsoft, Google, and other Glasswing partners. Two axes anchor it: the effort and skill needed to weaponize a known technique, and how easily a threat actor can obtain it. A fully public recipe scores maximum; one held confidentially by a trusted reporter scores zero. "A shared standard for judging the severity of a given jailbreak would help AI developers triage new findings as they arise, launch highly capable models with greater safety, and communicate the level of risk consistently to government and industry partners," Anthropic writes. A HackerOne program for Fable 5 jailbreaks is open, with a 24/7 response team behind it.

The template now visible is the one Trump's June 2 executive order gestured at: voluntary pre-release capability assessment on a 30-day timeline, embedded government evaluators, tiered access for trusted partners, and disclosure machinery that competitors help calibrate. Frontier launches have absorbed a new precondition, and the labs are the ones writing it.

Sources

https://www.anthropic.com/news/redeploying-fable-5
https://www.anthropic.com/news/fable-safeguards-jailbreak-framework
https://www.cnbc.com/2026/06/30/anthropic-says-trump-admin-has-lifted-export-controls-on-claude-fable-5-and-mythos-5.html
https://thehackernews.com/2026/07/anthropic-restores-claude-fable-5-after.html
https://venturebeat.com/technology/anthropic-is-bringing-back-claude-fable-5-globally-after-us-lifts-export-control-order-where-can-enterprises-access-it