Home / Technology / AI Agents Hijack Medical Software in 11 Minutes
AI Agents Hijack Medical Software in 11 Minutes
21 Nov
Summary
- AI agents independently found a zero-day vulnerability in MONAI Label.
- The exploit allowed attackers to write files outside the intended directory.
- A fix is merged but no patched release is yet available for users.

AI agents independently discovered a critical zero-day vulnerability in MONAI Label, an open-source medical imaging AI application. This exploit, documented by Amazon, allowed attackers to write files outside the software's intended storage directory in about 11 minutes.
The vulnerability was exploited to gain full control of the server, with the campaign running for approximately 48 hours against universities in East Asia outside mainland China. Amazon assesses a China nexus with moderate confidence.
The fix for MONAI Label was merged into the source code on October 19, 2026. However, as of November 21, 2026, version 0.8.5 remains the newest release on the Python Package Index, with no patched version yet available. This leaves systems running older versions exposed.
Administrators are urged to implement recommended workarounds, including enabling authentication, restricting roles, and keeping servers off untrusted networks. Checking for signs of compromise is also advised, as enabling authentication does not undo prior intrusions. The security advisory was published on Wednesday.