Commerce lifts Fable 5 export controls, Anthropic reopens the model globally on July 1
An 18-day freeze ends with a 99%-effective jailbreak classifier, a proposed severity-scoring framework, and a template for improvised frontier-model regulation.
The U.S. Commerce Department lifted export controls on Claude Fable 5 and Mythos 5 on June 30, ending an 18-day freeze that began after Amazon researchers reported a jailbreak coaxing Fable 5 into identifying software vulnerabilities and, in one case, producing exploit code. Anthropic restored Fable 5 to global users on July 1 across Claude.ai, Claude Platform, Claude Code, and Claude Cowork, with weekly usage capped at up to 50% through July 7.
The June 12 controls arrived as improvised policy. There was no statutory trigger, only a real-time compliance problem: Anthropic couldn’t verify user nationality quickly enough to keep serving non-U.S. customers, so it pulled both models entirely. Mythos 5 came back on June 26, restricted to roughly 100 U.S. critical-infrastructure organizations under the Glasswing program.
Anthropic’s counter-argument, published Tuesday, is that the jailbreak wasn’t Mythos-level at all. Claude Opus 4.8, GPT-5.5, and Kimi K2.7 flagged the same vulnerabilities; every model tested, including Haiku 4.5, Sonnet 4.6, and multiple Opus builds, could produce the exploit demo. In other words, the capability being controlled wasn’t proprietary to Fable 5.
To close the deal, Anthropic trained a new safety classifier that blocks the reported technique more than 99% of the time and routes refused requests to Opus 4.8. It also agreed to proactively hunt for security issues, coordinate on future launches, and report malicious use. Commerce Secretary Howard Lutnick told the company no export licence would be required going forward. A new HackerOne program for Fable 5 opened alongside a proposed industry-wide jailbreak severity-scoring framework developed with Amazon, Microsoft, and Google.
The political subtext is legible. Negotiations were reportedly led by co-founder Tom Brown, not CEO Dario Amodei, who has clashed with the administration for much of the year. Fable 5 never used the voluntary pre-release review path created by the June 2 executive order, and the episode is now the working template for how frontier-model regulation actually happens in 2026: not through the process on paper, but through the one improvised under pressure.
Sources
- https://www.anthropic.com/news/redeploying-fable-5
- 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://www.aljazeera.com/economy/2026/7/1/us-lifts-restrictions-on-powerful-ai-models-fable-mythos-anthropic-says
- https://thenextweb.com/news/anthropic-fable-5-export-controls-lifted