Anthropic restores its flagship AI after adding protections that satisfied US regulatorsLast updated: July 03, 2026 | 10:202 MIN READFor Anthropic, the redeployment marks the end of an extraordinary chapter.Supplied Most AI launches follow a familiar script. A company unveils a faster model, publishes benchmark scores and moves on to the next release.Claude Fable 5 didn't get that luxury.Just three days after Anthropic introduced the model, it vanished. Not because of a technical glitch or overloaded servers, but because the US government imposed emergency export controls after concerns emerged over how the model could be used in cybersecurity. Now, after weeks of negotiations and new safety measures, Fable 5 is returning to users.Get updated faster and for FREE: Download the Gulf News app now - simply click here.For Anthropic, the redeployment marks the end of an extraordinary chapter. For the AI industry, it may mark the beginning of a new one.The company says Claude Fable 5 is once again available through Claude.ai, Claude Code, Claude Cowork and its API after Washington lifted the restrictions on June 30. Access through cloud providers is being restored separately.The underlying model hasn't fundamentally changed.What has changed is the layer wrapped around it.Anthropic says it built a new classifier that watches for prompts aimed at uncovering software vulnerabilities or generating exploit code. When those requests are detected, they are redirected to the older Claude Opus 4.8 model instead. According to the company, the safeguard blocks the reported bypass technique in more than 99% of tests, though it also acknowledges that some legitimate coding requests may now be flagged by mistake.The episode offers a glimpse of where frontier AI is heading.When Fable 5 debuted in early June, Anthropic described it as the first publicly available Mythos-class model—a system built for demanding software engineering, long-running reasoning and autonomous knowledge work, while relying on external safety mechanisms to limit misuse. Its restricted sibling, Claude Mythos 5, remains available only to approved organisations through Project Glasswing.What made headlines, though, wasn't the model's benchmark performance.It was the government's response.According to Anthropic, the export controls followed a report from Amazon researchers showing a way to bypass one of Fable 5's cybersecurity safeguards. The company says its own testing found that several other advanced AI models could produce similar results, but it nevertheless worked with US officials to develop additional protections before the restrictions were lifted.The incident has sparked a wider debate across the AI industry. Developers have questioned whether regulating individual models is sustainable when competing systems offer comparable capabilities, while cybersecurity experts argue that stronger oversight is becoming inevitable as AI grows more capable.Anthropic has also announced a HackerOne bug bounty programme to encourage researchers to report jailbreaks and other weaknesses before they can be exploited. The company says it will continue working with government agencies as it prepares future frontier models for release.Also In This PackageGet Updates on Topics You ChooseUp Next