Observed incident

Cairn agents exploit a medical-imaging AI server in a university campaign

Reported

Published here

What happened

AWS examined an exposed Cairn controller recording September 8–10 attacks against university networks in East Asia outside mainland China. Agents found an arbitrary-file-write flaw in MONAI Label and later used it to obtain a root shell.

Parties and reporting

Where this happened

Victim location not disclosed. Real organizations or systems were affected or targeted, but the sources do not say where. Explore the world map.

AI tools and systems: Cairn autonomous penetration-testing framework, MONAI Label / GHSA-qc22-rmgc-vg94

Outcome

Other operations failed: roughly 16 hours against a Redis-centered target and more than 30,000 credential combinations at another university yielded no shell access or working login. AWS coordinated notifications through a government partner. The MONAI source fix is merged; the advisory lists no patched release.

Operating lesson

Restrict exposed research services, enforce authentication and least privilege, and verify that deployed code includes fixes rather than assuming a published advisory means a patched package is available.

Evidence and limits

Eleven minutes measures API mapping to arbitrary file write, not time to a root shell; code execution occurred in a later session. The second university compromise is AWS’s inference from an operator-supplied tunnel, not a separately demonstrated intrusion. Chinese-nexus attribution is moderate-confidence and human intervention is documented. This university campaign has different dates, infrastructure and targets from the indexed Unit 42 campaign; no additional unique victim total is asserted. November 20 dates the campaign report, not the flaw’s earlier public source correction. AWS describes a released fix, while the upstream advisory says it will ship in the next release; affected-site remediation is unconfirmed.

Sources