Senior Anthropic engineers sat down with Commerce Department officials and the National Cyber Director’s office in Washington on Monday, the first face-to-face meeting since Howard Lutnick’s June 12 letter forced the company to take Claude Fable 5 and Mythos 5 offline worldwide. According to a Trump administration official cited by The Globe and Mail, the parties broke without an agreement.
The letter arrived at 5:21 p.m. ET on Friday. According to a copy seen by Bloomberg News, Lutnick warned Dario Amodei that government permission would now be required before exporting either model to any destination, or to any foreign national regardless of location, and threatened criminal and civil penalties. Anthropic disabled both models the same day. Fable 5 had been live for three days; it launched June 9.
The legal scaffolding is itself the story. An export control expert cited by The Globe and Mail says Commerce invoked the Export Control Reform Act of 2018, and that this is the first time the department has reached for that authority. Other experts have questioned whether Commerce even has standing, because models delivered through remote access don’t fit cleanly inside an export regime built for hardware and code transfers.
The chain of events, per NBC News, began with a passing reference on a call between Amazon CEO Andy Jassy and Trump administration officials, citing a security finding on Fable 5. Foreign Policy reports Amazon raised the concern with Anthropic first, then escalated to Washington. Amazon is both a major Anthropic investor and a key cloud provider to the federal government, a dual position that did most of the work here.
Anthropic’s public statement reframes the finding sharply. The company says the government’s concern centers on a jailbreak demonstrated verbally, which it characterized as a “narrow, non-universal” method that surfaced “previously known, minor vulnerabilities.” The technique, per Anthropic, essentially involved asking the model to read a codebase and fix software flaws. The company notes that publicly available models, including OpenAI’s ChatGPT, can find the same flaws with no bypass at all. Anthropic also says it red-teamed Fable’s safeguards with the U.S. government, the UK AISI, and outside organizations for thousands of hours before launch, and received government approval to deploy.
By Sunday, more than 80 cybersecurity executives had signed an open letter to Lutnick and Cairncross backing Anthropic’s position. The Globe and Mail has separately reported that Anthropic previously refused to allow the U.S. military to use its models for domestic surveillance and fully autonomous weapons, after which the government placed the company on a national security blacklist. That history is the unspoken text on the table in Washington.
Amodei and Lutnick are both due in Evian-les-Bains for G7 meetings this week and may speak as negotiations continue. The first use of a 2018 export statute against a frontier lab is now precedent, regardless of how the Fable 5 file closes.
Sources
- https://www.bloomberg.com/news/articles/2026-06-16/read-the-lutnick-letter-that-led-anthropic-to-disable-mythos
- https://www.nbcnews.com/tech/security/anthropic-fable-5-ai-offline-trump-order-administration-claude-rcna350117
- https://www.theglobeandmail.com/business/article-anthropic-trump-officials-deal-restore-fable-5-mythos-5/
- https://foreignpolicy.com/2026/06/16/anthropic-trump-fight-mythos-fable-5-hegseth/
- https://www.anthropic.com/news/fable-mythos-access