Anthropic's two most capable models, Fable 5 and Mythos 5, remained globally disabled on Tuesday after a working-level meeting between senior Anthropic technical staff and Commerce Department officials in Washington ended without an agreement. National Cyber Director Sean Cairncross joined the session, per Reuters. Five days have now passed since the Commerce directive that pulled both products offline, and the legal vehicle behind it, the 2018 Export Control Reform Act, has never before been invoked against a frontier model.

The sequence is unusually compressed. Fable 5 launched publicly on June 9. By June 12, Amazon researchers had found a jailbreak trigger, something close to a "fix this code" prompt, that Anthropic says is reproducible on OpenAI's GPT-5.5. Amazon CEO Andy Jassy escalated the finding to Commerce Secretary Howard Lutnick, Treasury Secretary Scott Bessent, and former National Cyber Director Harry Coker Jr. By 5:21 p.m. ET that evening, Anthropic had received the directive. By midnight, both Fable 5 and Mythos 5 were dark for every customer worldwide.

Bloomberg published the Lutnick letter on Monday. Addressed to CEO Dario Amodei, it requires a license for any export of either model and threatens "prompt criminal and civil penalties" for non-compliance. The Washington Post, citing Semafor, reported that officials suspect a China-linked group accessed Mythos before the shutdown.

Anthropic's public statement pushed back hard. "We disagree that the finding of a narrow potential jailbreak should be cause for recalling a commercial model deployed to hundreds of millions of people," the company wrote. "If this standard was applied across the industry, we believe it would essentially halt all new model deployments for all frontier model providers."

That argument has industry backing. On Sunday, more than 80 cybersecurity executives signed an open letter to Lutnick and Cairncross supporting Anthropic's position. A person close to the company told Reuters that Anthropic's technical staff have been meeting with officials virtually every day since June 12.

The competitive cost is already visible offshore. On June 13, Zhipu AI launched GLM-5.2 and cited the U.S. ban as evidence of American unreliability, a framing the Financial Times amplified and Foreign Policy picked up.

There's a structural read here that the 2017–2019 Huawei entity-listing cycle made familiar: export-control authority, once stretched to cover a new technology class, tends to stay stretched. The Commerce Department now has a working precedent for treating a deployed commercial model as a controlled item on the strength of a three-day-old vulnerability report from a cloud partner with a financial stake. Whatever Anthropic negotiates back into production, that precedent doesn't unship. Amodei's company spent years cultivating itself as the safety-forward lab Washington could trust. The trust arrived, and it arrived with a license requirement.

Sources

https://www.anthropic.com/news/fable-mythos-access
https://www.bloomberg.com/news/articles/2026-06-16/read-the-lutnick-letter-that-led-anthropic-to-disable-mythos
https://www.theglobeandmail.com/business/article-anthropic-trump-officials-deal-restore-fable-5-mythos-5/
https://www.washingtonpost.com/technology/2026/06/15/how-anthropic-lost-white-houses-trust-then-its-flagship-product/
https://foreignpolicy.com/2026/06/16/anthropic-trump-fight-mythos-fable-5-hegseth/