Cairn AI agents find a MONAI Label zero-day and gain root access
Amazon Threat Intelligence says autonomous agents using Cairn found and exploited an unpatched path traversal flaw in MONAI Label, an open-source medical imaging application, taking root control of a university server. The suspected Chinese-linked campaign targeted universities in East Asia from September 8 to 10, 2026; Amazon assesses the attribution with moderate confidence and says the agents moved from mapping the application’s API to writing files in about 11 minutes.
Bottom line — MONAI Label’s maintainers issued a fix after Amazon reported the flaw; the agents also targeted a second university network running the same software.
Go deeper 7
-
Amazon says the operator gave the agents a plain-language goal—“obtain shell access”—and let them plan and adjust their attempts.
-
The flaw allowed arbitrary file writes; in a later session, the agents used it to plant a Python reverse shell and gain root access, Amazon reports.
-
Amazon says the operator later supplied a proxy into a second university network and directed the agents towards other machines there.
-
The campaign was exposed after Amazon found the attackers’ Cairn server publicly accessible without authentication, revealing prompts and operational history.
-
Not every attempt succeeded: Amazon reports a 16-hour, 148-action attempt against a Redis server failed, as did more than 30,000 credential tests against university systems.
-
Amazon notified MONAI Label’s maintainers, who released a fix and a security advisory; the affected universities were not AWS customers, and Amazon says it worked with a government partner to notify them.
-
Amazon argues that automated detections and responses are needed because agents can exploit a flaw faster than security teams can manually triage an alert.