On June 30, Anthropic announced that it would begin redeploying its latest model, Claude Fable 5, starting July 1[1]. The move follows the lifting of the US government's export controls on the model, making it available again to users worldwide on the Claude Platform, Claude.ai, Claude Code, and Claude Cowork[1]. Japanese media have also reported on the model's return after the controls were lifted[2].

From Export Controls to Redeployment

Fable 5 and its higher-tier counterpart, Mythos 5, were released on June 9, but on June 12 the US government applied export controls to both models[1]. Because the restriction on foreign nationals took effect immediately and there was no reliable way to verify nationality in real time, Anthropic suspended access to both models for all users[1]. The directive followed a report in which Amazon researchers had found a method of bypassing Fable 5's safeguards, prompting it to identify software vulnerabilities and, in one case, to produce code demonstrating how a vulnerability could be exploited[1]. However, Anthropic's own testing confirmed that less capable existing models, including Claude Opus 4.8, GPT-5.5, and Kimi K2.7, could identify the same vulnerabilities and produce the same demonstration[1].

A New Safeguard That Blocks Over 99% of Attempts

For the redeployment, Anthropic trained an improved safety classifier that targets the reported bypass technique[1]. The company says the classifier now blocks the specific technique in more than 99% of cases[1]. Users are notified when a request is blocked, and the request is instead routed to Claude Opus 4.8[1]. Researchers at the US Department of Commerce's Center for AI Standards and Innovation (CAISI) tested both the prior and new safeguards and agreed that they are extraordinarily strong[1]. At the same time, the stronger classifier may flag more benign requests during routine coding and debugging, and Anthropic says it will keep refining it[1].

Access Terms and an Industry-Wide Jailbreak Framework

For Pro, Max, Team, and select Enterprise plans, Fable 5 is included for up to 50% of weekly usage limits through July 7, after which it will be available via usage credits[1]. Access on AWS, Google Cloud, and Microsoft Foundry will be re-enabled in stages[1]. Anthropic also revealed that, together with Amazon, Microsoft, Google, and other Glasswing partners, it is developing an industry-wide framework for assessing the severity of AI jailbreaks (techniques that bypass a model's safeguards)[1]. The proposal would score a jailbreak on four criteria: capability gain, breadth of that gain, ease of weaponization, and discoverability[1].

Summary

Just about three weeks after its suspension under export controls, Claude Fable 5 has returned with stronger safeguards. This sequence of events stands out as a case study in balancing the safety and availability of frontier AI models, and in how government and industry can collaborate.

Source: https://www.anthropic.com/news/redeploying-fable-5

Source: https://www.watch.impress.co.jp/docs/news/2121360.html