Swap “censorship” for “unban”? Fable 5 returns tomorrow, but it’s no longer the original Fable 5.

After weeks of restrictions, Anthropic's Fable 5 has finally been approved to relaunch, but at the cost of its core capabilities being significantly tightened.

On Tuesday, June 30 local time, the US Department of Commerce officially lifted the export restrictions on the Anthropic Fable 5 and Mythos 5 models. Anthropic subsequently confirmed on the X platform that access would be restored on Wednesday.

However, Anthropic posted on X that compared to previous versions, the relaunched Fable 5 incorporates stricter safety classifiers. Certain programming and debugging requests will be automatically downgraded to the older Opus 4.8 model, meaning scenarios where users can actually utilize Fable 5's capabilities are significantly reduced.

This change quickly sparked heated discussion in the developer community. Many industry professionals believe that although Fable 5 is nominally reopened, its most competitive capabilities have been "cut," making it more like a strictly limited version rather than the originally released full model.

Fable 5 is back, but its core abilities "have been downgraded"

The company announced a major adjustment: the new version of Fable 5 adds a safety classifier to identify requests that may involve sensitive cybersecurity capabilities. Anthropic admits that the cost of this classifier is "more frequent misidentification" of normal requests, including everyday programming and code debugging tasks.

The official statement says:

The new classifier also means that more normal requests will be mistakenly identified during daily coding and debugging tasks.

For these identified requests, the system will not invoke Fable 5, but will automatically switch to the previous generation Opus 4.8 model for completion. Developers generally believe that this means Fable 5's most powerful code generation capability can no longer be fully accessed.

Developer Wise commented: Fable 5 is no longer Fable 5. To lift the export restriction, Anthropic added new safety classifiers, while handing over a large number of routine programming and debugging tasks back to Opus 4.8.

Renowned crypto market analyst Miles Deutscher also pointed out that the new Fable 5 not only restricts coding capabilities, but also only allows users to use the model within about 50% of the quota, while remaining requests still fall back to the old model. Therefore, this "restart" seems more like a restricted reopening rather than a true restoration of its original abilities.

From total lockdown to conditional unblocking

This adjustment is the result of several weeks of negotiations between the US government and Anthropic.

On June 12, the US Department of Commerce, citing national cybersecurity risks, prohibited Anthropic from providing the Fable and Mythos advanced models to users. The regulator's biggest concern focused on the so-called "jailbreak" risk—that users could bypass model safety restrictions and obtain stronger cyber-attack capabilities.

Subsequently, the US government last week allowed limited access to Mythos 5 for certain US companies and government agencies, while the consumer-grade Fable 5 remained blocked. According to previous media reports, to push for lifting the ban, Anthropic ultimately adjusted its negotiation strategy, no longer insisting that the "jailbreak risk cannot be fully eliminated," but instead promised to implement stricter safety mechanisms to reduce the probability of risk.

The new safety classifier added to the relaunched Fable 5 is precisely the main outcome of this compromise between both sides.

AI regulation enters the "capabilities available, but must be controlled" phase

The relaunch of Fable 5 further reflects a shift in the US approach to AI regulation.

Instead of directly prohibiting model deployment, the US government prefers to require companies to implement safety classifiers, restrict sensitive features, and dynamically downgrade models to allow ongoing oversight of advanced models.

Anthropic is not the only affected company. Previously, OpenAI's latest model, ChatGPT-5.6, was similarly only available to a small group of users under White House and regulatory agency demands, with its release schedule clearly slowed. OpenAI management has publicly expressed a desire to quickly establish a clear and stable AI model evaluation and approval mechanism to reduce uncertainty caused by temporary regulation.

For the entire AI industry, Fable 5's "lifting of restrictions" may signal a new regulatory paradigm is emerging: advanced models can be released, but the most sensitive and most powerful abilities will increasingly be dynamically managed through safety classifiers, ability downgrades, and access restrictions. Developers will have a usable Fable 5, but it may no longer be the complete Fable 5 of the beginning.

Risk warning and disclaimerThe market has risks, and investment needs caution. This article does not constitute personal investment advice, nor does it take into account the specific investment objectives, financial situation, or needs of any particular user. Users should consider whether any opinions, viewpoints, or conclusions in this article fit their particular circumstances. Investment based on this, responsibility is your own.