In brief
Anthropic has halted its newest AI model, Claude Fable 5, after U.S. authorities raised security concerns. The order requires the company to suspend foreign nationals from using the tool because of a potential jailbreak method. Despite no detailed vulnerabilities being named, researchers warn that the model could help bypass safety restrictions. The move occurs as Anthropic faces a Pentagon lawsuit claiming the tool should not be used by government agencies.
Anthropic has pulled its latest model Claude Fable 5 after American authorities flagged security concerns hours after it went live.
The company said it had to suspend foreign nationals from using the tool because the government identified a possible bypass method.
Although no specific vulnerability was named, researchers say the model could be used to jailbreak built‑in safety mechanisms.
The decision comes amid a lawsuit in which Anthropic argues that a Pentagon order to bar the tool from government use is unenforced.
Critics warn that halting the model may slow AI progress and estrange governments from the technology.
Experts say the new generation of large language models can pose serious cyber threats, yet still the industry stresses the need for safety protocols.
