AI Model Report

Reviews · JULY 4, 2026

Fable 5 returns after 19-day export-control blackout, with a new classifier and a government review deal

The Commerce Department lifted its June 12 export-control directive on June 30. Anthropic restored Fable 5 worldwide July 1 alongside a new jailbreak classifier, a HackerOne program, and a commitment to pre-release government access for future frontier models.

By Karl Strauchman · Senior model reviewer · July 4, 2026

Fable 5 came back online globally on July 1, 2026 at 3:31pm ET, ending a 19-day blackout that began when Commerce Secretary Howard Lutnick's June 12 export-control directive forced Anthropic to disable both Fable 5 and its Mythos 5 cybersecurity sibling for every customer worldwide. The directive landed at 5:21pm ET; the models went dark shortly after. What returns today isn't just the product. It's a new operating template for what happens when the federal government treats a jailbreak like a munitions leak.

The precipitating event was narrow. Amazon researchers surfaced a prompt that, by Anthropic's own account, coaxed Fable 5 into flagging "a small number of previously known, minor vulnerabilities" and, in one case, writing code demonstrating abuse. The Commerce response wasn't narrow. The original directive required cutoff for any foreign national inside or outside the United States, including Anthropic's own non-citizen staff. Anthropic had no per-request way to verify nationality, so it pulled the models entirely.

The blackout is where the interesting story lives. During the standoff, Anthropic warned that the precedent would "essentially halt all new model deployments for all frontier model providers." Multiple executives argued, on background, that the three-week freeze was handing runway to cheap, capable Chinese open-weights competitors. Per WIRED's reporting, CEO Dario Amodei was replaced in meetings by co-founder Tom Brown, and the posture shifted from arguing that no provider can guarantee zero jailbreaks to committing to concrete safeguards. Lutnick's June 30 withdrawal letter was addressed to Brown.

Mythos 5 had already been partially restored to trusted U.S. organizations on June 26, four days before the broader license requirement was withdrawn. That sequencing matters. It's the shape of a negotiated settlement, not a policy reversal.

The concessions are structural. Anthropic is standing up a 24/7 triage team, has opened a HackerOne bug-bounty program, and commits to deploying highest-severity fixes "the moment severity is confirmed." The Glasswing cybersecurity program, with Microsoft and Google as partners, sits alongside a new jailbreak classifier. And per President Trump's June 2 executive order, future frontier models will move through a pre-release government access channel before shipping.

For customers: AWS, Google Cloud, and Microsoft Foundry are being re-enabled "as quickly as possible." Pro, Max, and Team subscribers get up to 50% of their weekly tier allowance for Fable 5 usage at no added cost through July 7, after which the model moves to usage credits. Standard Enterprise seats get no included allowance; all usage bills through credits.

The 2019 Huawei entity-listing established that a single Commerce action could rewire a global supply chain overnight. June 12 extended the logic to model weights. What Anthropic negotiated back isn't deregulation. It's a permission structure, with the classifier as the tollbooth.

Sources

  • https://www.anthropic.com/news/fable-mythos-access
  • https://www.cnbc.com/2026/06/30/anthropic-says-trump-admin-has-lifted-export-controls-on-claude-fable-5-and-mythos-5.html
  • https://www.nbcnews.com/tech/tech-news/anthropic-suspends-new-ai-models-fable-mythos-government-directive-rcna349901
  • 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