Anthropic's Claude Mythos AI model is uncovering critical vulnerabilities in software at unprecedented scale, prompting cybersecurity chiefs to demand stronger coordination between government and businesses protecting hospitals, banks and utilities. Companies given early access to the tool say joint action across the public and private sectors is essential.

Cisco president Jeetu Patel described the situation as moving from "a world pre-Mythos to a world post-Mythos." Fifth Third Bank has received nearly 150 software updates from Microsoft since Mythos's release. Palo Alto Networks warns the technology could enable hackers to develop "autonomous attack agents" that chain together vulnerabilities to bypass security systems.

The volume of vulnerabilities identified could trigger a "flood of patches," challenging businesses that cannot afford system downtime. Critical infrastructure groups are especially attractive targets since they often run older software. This week, Anthropic said it was investigating reports that unauthorized users had gained access to Mythos through third parties.

Central bankers and regulators have demanded expedited access to Mythos, but the startup has declined to provide a timeline.