AI Agents Became a Security Liability

AI systems moved from a liability question to an active source of security incidents, with agents interacting with public sites and AI-linked attacks reaching banks. The week also exposed a defensive mismatch: leaders called AI attacks the threat they were least prepared to handle even as AI tools found hidden patient-record access and offered open-source security scanning.

The South Korean bank case connected recurring breach concerns to a still-unsettled question about how AI was used. Separately, arrests widened scrutiny of both ShinyHunters operators and the commercial ecosystem around ransomware response.