Anthropic, White House end Monday talks with no deal on Fable 5 and Mythos 5 ban
Senior Anthropic staff met Trump administration officials in Washington on Monday to try to lift Friday's export control order disabling Claude Fable 5 and Mythos 5 for every customer worldwide. No agreement was reached.
Senior Anthropic staff left Washington on Monday with nothing to show after a day of talks with Trump administration officials aimed at reversing Friday’s export control directive, which forced the company to disable Claude Fable 5 and Mythos 5 for every customer worldwide. People familiar with the talks told CNBC and Foreign Policy that no resolution was reached. Neither side responded to Foreign Policy’s request for comment on what was discussed.
The sequence is unusually compressed. Fable 5 and Mythos 5 launched the previous Tuesday. By Friday at 1:00 p.m. ET a verbal call came through instructing Anthropic to disable the models, followed at 5:21 p.m. ET by a formal letter citing “national security authorities.” Five business days from product unveil to global takedown is, even by 2026 standards, a new tempo.
The proximate cause, per PBS NewsHour, is that Amazon CEO Andy Jassy flagged the concern to White House officials on Friday after Amazon researchers found a method of getting through Fable 5’s guardrails. Commerce Secretary Howard Lutnick’s letter, as reported by The Hill citing Reuters, raised the prospect that the models could be used by military intelligence in countries including China and Russia.
Anthropic disagrees, sharply. Its Friday statement calls the underlying issue a “potential narrow, non-universal jailbreak” reflecting “previously known, minor vulnerabilities,” and argues that “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 notes that other public models, including OpenAI’s GPT-5.5, can produce the same vulnerabilities. It frames the standoff as a “misunderstanding” and says it’s working to restore access “as soon as possible.”
Luta Security CEO Katie Moussouris, who reviewed the third-party paper underlying the order, told The Hill that researchers had used open-source code with known vulnerabilities, and that Fable initially refused before a manual, multi-step bypass succeeded. An open letter signed by security staff from Nvidia, Zoom, Mercedes-Benz, and former U.S. government and Google personnel has urged Lutnick to lift the controls.
The 2019 Huawei entity-listing established that export-control machinery can be turned on a single American counterparty’s product line within a news cycle. Friday was the first time it was turned on a frontier model.
Sources
- https://www.anthropic.com/news/fable-mythos-access
- https://foreignpolicy.com/2026/06/16/anthropic-trump-fight-mythos-fable-5-hegseth/
- https://www.cnbc.com/2026/06/15/anthropic-mythos-trump-ai.html
- https://thehill.com/policy/technology/5926417-anthropic-fable-mythos-ai/
- https://www.pbs.org/newshour/show/anthropic-disables-new-ai-model-after-white-house-security-directive
— END —