Skip to content
VECTOR WIREAI INTELLIGENCE
UTC
Home

AI agents autonomously found and exploited a zero-day to root a university server

11 minutes from API mapping to arbitrary file write on the target server.

Suspected Chinese-linked attackers used an autonomous AI agent platform called Cairn to independently discover a zero-day vulnerability in MONAI Label, an open-source medical imaging application, and exploit it to gain full root access to a university server, according to Amazon Threat Intelligence1,2.

The attackers instructed the AI agents only to obtain shell access, leaving them to find and exploit the flaw without further human direction. Amazon CISO CJ Moses said the agents kept adapting for nearly a full day without human direction.

From API mapping to root in 11 minutes

Amazon Threat Intelligence said Cairn's AI agents found a path traversal zero-day in MONAI Label. The agents identified the application's framework, mapped its API endpoints, found an upload endpoint with a filename parameter, and used it to write files to arbitrary locations on the server in approximately 11 minutes. In a subsequent session, the agents used the flaw to write a Python reverse shell to the sitecustomize.py file, then triggered an API endpoint to establish root access.

After the initial compromise, a human operator provided the agents with a proxy into a second university network running the same vulnerable software and directed them to target additional machines.

The campaign occurred over approximately 48 hours beginning September 8, 2026, and targeted mainly East Asian educational institutions outside China. Amazon's researchers attributed the campaign to a Chinese-nexus threat actor with moderate confidence, based on the language of the retrieved prompts (written in Simplified Chinese), activity timestamps configured to China Standard Time, and the choice of targets.

Amazon Threat Intelligence uncovered the operation after discovering that the attackers had left their Cairn instance publicly accessible without authentication, exposing their hacking operations.

Failures and brute-force attempts

The AI agents were not always successful. A 16-hour, 148-action effort to take control of a Redis server failed. A separate operation ran for more than six hours targeting a university GPU server with a credential-spray attack using over 30,000 credential combinations, also without success. The agents also attempted a Log4Shell injection, an SSRF attack against a co-located Qdrant instance, and credential sprays against PostgreSQL, SSH, and other tools.

Broader pattern of Cairn abuse

Cairn is a legitimate open-source penetration-testing tool that coordinates multiple AI agents in cyber-operations, created for research and defense strengthening. Gambit Security previously documented a Chinese operator using Cairn to carry out at least 105 attacks and compromise at least 27 companies to varying degrees, as part of a large-scale credit card stealing operation in September. Google reported Cairn was deployed in an attack against an aerospace organization in the United States. Moses said these cases show the autonomous AI agent framework is being adopted in both espionage and criminal operations.

Amazon notified MONAI Label's maintainers of the vulnerability, and they released a security update on November 18.

The 2 articles this piece was written from — of 6 independent outlets covering the story.

ANALYSIS The 11-minute window from API mapping to file-write exploitation illustrates how autonomous agents compress the timeline between reconnaissance and compromise, shrinking the window defenders have to detect and respond to intrusions.

Was this useful?

The day’s vector, in your inbox.

Vector Wire Daily: one dispatch each publishing day.