AI & Tech

China attackers deployed AI agents that autonomously found and exploited a 0-day to take over a university server

Suspected Chinese spies deployed AI agents which worked autonomously to find a zero-day in an open source application and then abuse the flaw to break into university networks, gaining full control of a server, according to Amazon CISO CJ Moses.

✦ Extractive summary — built from the sources' own reporting•From 1 source•2 min read•Updated November 20, 2026, 5:30 PM UTC
Key points
  • Before the AI agents found and exploited the zero day, however, they tried to break into another university’s Redis server over about 16 hours and 148 sequential actions, including abusing Redis's replication feature to load unauthorized code, escaping its scripting sandbox, writing to its configuration files, and corrupting its memory.
  • On September 10, Amazon's Threat Intelligence Team found a server with a publicly hosted instance of Cairn that provided visibility into the campaign targeting universities in East Asia, outside Mainland China, Moses said in a Friday report shared with The Register.
  • Meanwhile, a separate operation running for more than 6 hours targeted a university GPU server with another credential-spray attack, compiling a list of 80 real user names found across the school’s public websites, academic papers and other documents.

In another session, after the agents discovered the zero-day, the human operator provided them with a proxy that tunneled into a second university network, and connected to a server running the same vulnerable code.

Additional details
  • In this case, the agents found and exploited a zero-day MONAI Label, an open-source medical imaging AI application.
  • In about 11 minutes, the agents went from mapping the application’s API to arbitrarily writing files on the server." Later, the agents abused the same flaw to execute a Python reverse shell running as root.
  • None of these were successful, and when those actions failed, the agents also attempted a Log4Shell injection against a different service, a server-side request forgery (SSRF) attack against a vulnerability in a co-located Qdrant instance, and several credential sprays against PostgreSQL, SSH, and other tools.
  • "The agents identified the application's framework, mapped its API endpoints, and found an upload endpoint with a filename parameter," Moses wrote.
  • "Neither operation succeeded, yet together they show that the agents kept adapting for nearly a full day without human direction," Moses wrote.

How this summary was made: extractive summary — built from the sources' own reporting. It is drawn only from the reporting of the sources listed below — read the originals for full detail. How we summarize →

Advertisement

Every article behind this summary — 1 article from 1 outlet