Related: The AI IPO race heats up as Anthropic and OpenAI eye Wall Street.
The US government has lifted the export controls that forced Anthropic to switch off its most capable AI models, clearing the way for Claude Fable 5 to return to users around the world. In a notice to the company, the Commerce Department’s Bureau of Industry and Security withdrew the restrictions it had placed on Fable 5 and its sibling Mythos 5, and Anthropic said it would begin restoring access the next day. The move ends a weeks long standoff that had taken two of the most powerful commercial models offline and rattled the many businesses that had built on them.
The dispute began in June, when the government issued an export control directive that suspended all access to Fable 5 and Mythos 5 by any foreign national, whether inside or outside the United States, a definition sweeping enough to include Anthropic’s own foreign national employees. The order was grounded in national security authorities, and while officials did not spell out their reasoning in public, reporting indicated they were worried that Fable 5 could be jailbroken to unlock the cybersecurity capabilities of Mythos 5, Anthropic’s strongest model in that domain. Faced with a directive it could not easily comply with while keeping the models running, Anthropic shut both of them off.
The thaw came in stages. Last week the government allowed Anthropic to restore Mythos 5 for a select group of approved organizations that operate and defend critical infrastructure in the United States, a narrow reopening rather than a full reversal. The decision announced this week goes much further, removing the licensing requirements that had applied to exports, reexports, and certain transfers of the models, which is what allows Fable 5 to return to general availability rather than a walled off list of users.
The reversal came with conditions. According to the letter from Commerce Secretary Howard Lutnick, Anthropic no longer needs an export license after agreeing to a set of commitments, including to proactively detect and address security risks associated with the models, to work with the government on protocols for how future models are released, and to report any malicious activity it finds involving its systems. In effect the government traded a blunt shutdown for an ongoing arrangement in which Anthropic takes on more responsibility for policing how its most capable models are used and secured.
The episode is a marker of how tangled frontier AI and national security policy have become. A government reaching for export control authorities, tools built for physical goods and sensitive technologies, to govern access to a software model shows how officials are treating the most capable systems less like products and more like strategic assets. For Anthropic, the weeks of downtime were a costly reminder that a model can be regulated out of a customer’s hands with little warning, and that the terms of doing business at the frontier now include a direct relationship with the state.
For users, the practical result is simpler, Fable 5 is coming back. Anthropic thanked its customers for their patience and the officials who worked through the resolution, and said it would share more detail as access is restored. The larger questions the standoff raised do not disappear with the lift, though. The government has established that it will step in when it judges a model too risky to leave unrestricted, and it has set a template, security commitments and reporting in exchange for access, that is likely to shape how the next powerful model reaches the market. The immediate crisis is over, but the precedent it set will outlast it.
Related on Entrelligence: a Chinese open model matching Anthropic’s Mythos, the White House letting Anthropic restore its top model, and Anthropic’s drug discovery program for neglected diseases.
Your First 10 AI Skills
10 practical AI skills, copy-paste prompts and a 7-day plan to start using AI with confidence.
Download the guide →