Anthropic officially began redeploying its Fable 5 AI model globally on July 1, 2026. The Fable 5 Unban comes after the lifting of US export controls that previously forced the company to withdraw the model from the market. The return of Fable 5 marks the end of a brief but intense dispute between the AI developer and the US government over national security and software safety.
The controversy began on June 12, 2026, when the US government imposed immediate export controls on Fable 5. These rules required Anthropic to block all foreign nationals from using the model. Because Anthropic did not have a way to reliably verify the nationality of its users in real-time, the company felt it had no choice but to suspend access to the model entirely. The government’s decision was sparked by a “jailbreak” method discovered by researchers at Amazon. This exploit allowed the AI to bypass its built-in safety features to identify software vulnerabilities and, in some instances, generate code to exploit those flaws.
Anthropic initially pushed back against the ban, with some experts describing the government’s reaction as “significantly overblown”. The company pointed out that other leading models, including GPT-5.5 and its own Opus 4.8, were also capable of identifying the same vulnerabilities during testing. Anthropic said that the jailbreak did not reveal any unique, high-level capabilities that weren’t already present in other available AI systems.
To resolve the situation, Anthropic worked closely with the US government over a two-week period to analyze the model and strengthen its safeguards. The company developed a new safety classifier specifically designed to block the reported jailbreak behavior. Anthropic states that this fix is successful in over 99% of cases. Under the new system, if a user makes a request that is now blocked by these safeguards, the system automatically forwards the task to the older Opus 4.8 model. These updated security measures were eventually reviewed and endorsed by researchers from the Commerce Department.
The decision to lift the ban also comes amid growing concerns regarding global AI competition, particularly with China. Reports indicate that Chinese AI models are now performing as well as leading US models like Mythos in cybersecurity tasks, such as finding bugs. They often do this at a much lower cost.
Currently, Fable 5 is back online and accessible through the Claude Platform, Claude.ai, Claude Code, and Cowork. However, users should expect a slightly different experience, as the model will fall back to Opus 4.8 for certain coding and debugging tasks to ensure safety. Additionally, Anthropic has restored its Mythos 5 model for a select group of approved US organizations following government approval on June 26
