Anthropic's Mythos sparks security fears

Anthropic's latest AI model, Mythos, has triggered urgent discussions between the company and government officials over its powerful cybersecurity capabilities.

The firm has limited the model's release, claiming it can identify and exploit previously undiscovered vulnerabilities in major operating systems and web browsers.

European cyber agencies have expressed concern over being excluded from internal testing, while the US Treasury is seeking access to hunt for flaws in financial systems.

The debate highlights a growing divide between AI developers and regulators over the safe deployment of advanced 'reasoning' models.