Hackers Used Cairn AI Agents to Find a Zero-Day and Take Over a Server
Reported by BankInfoSecurity, Bleeping Computer, The Register and 2 more outlets5 sources ↓
Amazon says Cairn agents found and exploited a zero-day in MONAI Label, an open-source medical imaging application, and gained root access to a server. The campaign targeted East Asian universities outside mainland China.
Key points
- The MONAI Label flaw let attackers write files anywhere on a server and execute code as root
- Agents went from mapping the application’s API to arbitrary file writes in about 11 minutes
- Amazon attributed the campaign to a China-aligned actor with moderate confidence
- MONAI Label maintainers released a fix on November 18, according to Bleeping Computer
Sources
- BankInfoSecurityRead ↗Hackers Use Cairn to Hack Medical Imaging Application
- Bleeping ComputerRead ↗Hackers use autonomous AI tool Cairn to find zero-day, take over server
- The RegisterRead ↗China attackers deployed AI agents that autonomously found and exploited a 0-day to take over a university server
- ShiftDelete.NetRead ↗Hackers Used Cairn AI Agents to Find a Zero-Day and Take Over a Server
- news.lavx.huRead ↗AI agents find and exploit zero-day in university server, Amazon says
Details
Amazon found the campaign after locating a publicly accessible Cairn instance on September 10, which exposed its prompts and activity history. Bleeping Computer reports the activity ran from September 8 to 10 and mainly targeted educational institutions in East Asia outside mainland China. Amazon based its moderate-confidence attribution on Simplified Chinese prompts, China Standard Time timestamps and target selection.
The agents identified a path traversal flaw in MONAI Label’s upload endpoint, which allowed them to write files outside the intended directory. In a later session, they used the flaw to place a Python reverse-shell payload in a file loaded when Python starts, then triggered an API endpoint to gain root access. The operator also supplied a proxy into a second university network running the vulnerable software.
Other attempts failed. The articles report a 16-hour effort against a university Redis server that did not succeed; Bleeping Computer also says agents tried more than 30,000 credential combinations against a university GPU server over six hours without gaining access. Amazon declined to identify which AI models were used with Cairn, according to BankInfoSecurity.
Written from coverage as of
Images
- BankInfoSecurity
- Bleeping Computer
- news.lavx.hu
- Bleeping Computer
Stocks in this story
Change since open: 20 Nov to 21 Nov 2026
Last refresh