Amazon says suspected China-linked attackers used autonomous “AI agents” to discover and exploit a zero-day vulnerability in MONAI Label, an open-source medical-imaging server. The agents identify a flaw that lets an attacker write files outside the server’s intended storage directory, enabling root access on a university system.
Across the reports, the key technical focus is on how the vulnerability is reached and what it enables: unauthorized file writing beyond storage controls and subsequent compromise. Amazon Threat Intelligence attributes the activity to China-linked hackers and describes the defect affecting MONAI Label versions 0.8.5 and earlier in default configurations.
VentureBeat adds timing and release-status context, stating that exploitation occurred in about 11 minutes and that while a fix is merged into the source code, a patched release was not yet available. Other outlets emphasize the AI-agent discovery and the medical software target, with SecurityWeek also framing the issue as an AI-agent-driven exploitation path against medical software.