Anthropic announced that the artificial intelligence model Claude 5 Fable will once again be available to users worldwide starting this Wednesday, after a period of restrictions motivated by cybersecurity concerns raised by the United States government. The release comes after negotiations between the company and U.S. authorities, resulting in the implementation of new mechanisms to reduce the risk of misuse of the technology.
According to the company, the new version of Claude 5 Fable comes equipped with an updated set of safety classifiers, developed to identify and block requests related to cybersecurity activities considered sensitive. The measure is part of a commitment undertaken by Anthropic to meet the requirements discussed with the U.S. government.
During this initial return period, some routine tasks, such as programming and code debugging, may be automatically directed to the model Claude Opus 4.8 when the new filters identify possible risks. The company explained that this decision seeks to balance protection against abuse without completely compromising the user experience.
Anthropic also stated that it will continue refining the classifiers in the coming weeks to reduce false positives and improve the system's ability to differentiate legitimate requests from real attempts to exploit the AI.
In addition to reactivating the model, the company revealed that it has begun drafting a consensus framework aimed at assessing the severity of attacks known as "jailbreaks" in artificial intelligence models. The initiative is being developed in partnership with tech giants, including Amazon, Microsoft, Google, and other participants in the Glasswing alliance.
The goal is to create common criteria to measure risks and define how AI development companies should respond when vulnerabilities are discovered. Anthropic also invited other organizations and AI model providers to participate in the initiative.
As part of this new agreement, the company will expand its collaboration with the United States government in security testing and the development of safeguards for future models. The plan includes early access for authorities to new versions of the AI before public release, sharing information about exploitation attempts, and joint research to strengthen protection mechanisms.
When announcing the return of Claude 5 Fable, Anthropic thanked users for their patience during the period of unavailability and highlighted that the resumption of the service represents a joint effort among government, technology companies, and the research community to balance innovation and security in the advancement of artificial intelligence.