Anthropic Pulls New AI Tool After US Security Alarm Over Hacking Risks

The San Francisco AI company suspended access to Claude Fable 5 days after its public release following government warnings that the model posed cybersecurity and hacking dangers.

Anthropic suspended public access to Claude Fable 5 shortly after its release this week.
US government officials raised concerns that the tool could be exploited for hacking and cybersecurity attacks.
The suspension applies to new user access while the company reviews the model's risk profile.
Anthropic has not confirmed a timeline for reinstatement or what modifications may be required.
The move marks one of the first instances of a US government security directive prompting a major AI firm to pull a live product.

Anthropic suspended access to its newly released AI model Claude Fable 5 this week after US government officials raised alarms that the tool could be used to conduct cyberattacks and hacking operations, the BBC reported Wednesday.

The San Francisco-based company had released Claude Fable 5 to the public days earlier. Within that window, security concerns escalated to the point that Anthropic pulled new access to the model entirely.

Government officials warned that the model's capabilities presented risks significant enough to warrant immediate action, according to the BBC's account of the episode.

The suspension is one of the first on record in which a direct US security intervention forced a major AI developer to withdraw a live commercial product from public reach.

Anthropic has not issued a public timeline for when or whether Claude Fable 5 will return, or what changes the company would need to make to satisfy government requirements.

Congressional committees overseeing AI and cybersecurity policy are expected to press federal agencies for a full account of the directive in hearings scheduled for later this month.