Hundreds of millions of users lost access to Anthropic's most capable AI models Friday night — not because of a flaw serious enough to justify the move, according to Anthropic itself.
Anthropic disabled access to its Fable 5 and Mythos 5 models to comply with an export control directive from the US government that cited "national security authorities." The shutdown came less than a week after the models launched publicly, catching developers and enterprise customers completely off guard.
Commerce Secretary Howard Lutnick sent a letter to Anthropic CEO Dario Amodei saying the Mythos 5 and Fable 5 models would be subject to export controls to any location outside the US and to all foreign persons within the country.
An administration official told Axios the Commerce Department decided to take the action after another company claimed it was able to jailbreak Mythos, alarming the administration about possible national security risks. The administration had tried to get Anthropic to pause releasing the latest models but was unsuccessful.
Both models stemmed from Claude Mythos Preview, a highly advanced model intended for security research, which was capable of finding security bugs and flaws. Access to Mythos Preview was initially limited to a small group of companies and research partners through Project Glasswing.
In the weeks that followed, participants reported identifying and fixing numerous security issues with the model's help — Mozilla alone said it resolved hundreds of vulnerabilities as a direct result of using Mythos Preview.
What the Government Actually Found
Anthropic's public statement pushes back hard on the technical premise of the directive. The government's evidence, to date, consists only of a verbal claim of a potential narrow, non-universal jailbreak — essentially asking the model to read a specific codebase and fix any software flaws.
Anthropic reviewed a report it believes is the basis of the directive and validated that the level of capability displayed is widely available from other models, including OpenAI's GPT-5.5, and is used every day by the defenders who keep systems safe.
The distinction between a "universal" and "non-universal" jailbreak matters here. A universal jailbreak can very broadly bypass a model's safeguards, unblocking a wide range of capabilities. A non-universal jailbreak can elicit some information only in specific, narrow circumstances. Anthropic says no one — including government red-teamers — has found the former.
In the weeks leading up to the launch of Fable, Anthropic worked with the US government, the UK AISI, multiple private third-party organizations, and internal teams to red-team Fable's safeguards for thousands of hours in total. Those tests showed that Fable's safeguards are substantially more effective than those of any previously deployed model.
A Compliance Call Anthropic Openly Disagrees With
Anthropic is complying — but isn't staying quiet about it. The company argues the government's action sets a precedent that could freeze the entire AI industry. Anthropic says that if this standard — pulling a commercial model over the finding of a narrow potential jailbreak — were applied across the industry, it would essentially halt all new model deployments for all frontier model providers.
Anthropic adopted a defense in depth strategy with Fable 5, aiming to make jailbreaks either narrow or very expensive to produce, combined with thorough monitoring to quickly detect and shut down any successful attacks. This is also why the company required 30-day retention of customer data with Fable — a policy change that carries real costs with customers, but that allows research and mitigation of jailbreaks.
What Happens to Your Access Now
Across Claude products, new sessions will run on your selected default model or Opus 4.8, and existing Fable 5 sessions will end with an error. On the Claude Platform, requests to Fable 5 will also return an error — developers need to update their integrations to other Claude models. Every other Claude model remains fully accessible.
Anthropic says it will share more technical details within 24 hours and is actively working to restore access. The company described the shutdown as a misunderstanding and says it believes the directive does not meet the standard of being transparent, fair, or grounded in technical facts.
This is a developing story — check back for updates as Anthropic releases further documentation.