Suspected China-linked actors used AI agents to exploit a MONAI Label zero-day
Original title: China attackers deployed AI agents that autonomously found and exploited a 0-day to take over a university server
Amazon says the agents wrote files and launched a root Python reverse shell after failed attacks on other campus services.
Amazon CISO CJ Moses said suspected Chinese operators used the open-source Cairn framework so AI agents could find a path-traversal zero-day in MONAI Label and take control of a university server. Earlier attempts against Redis, a Log4Shell injection, a Qdrant SSRF flaw and credential sprays failed. Maintainers were notified and released a fix and advisory. A separate September report said Cairn was also used against at least 27 companies.
Unconfirmed: Amazon calls the operators suspected Chinese spies based on language, timestamps and a Cairn server, not a named group.
- The campaign ran about 48 hours beginning September 8, Amazon said.
- Prompts were in Simplified Chinese; script timestamps used China Standard Time.
- Redis attempts lasted about 16 hours and 148 actions and failed.
- A GPU-server spray compiled 80 names from public university sources.
- Gambit Security reported at least 105 Cairn attacks and 27 compromises.
Impact: root reverse shell and full control of a server
| Product | Affected |
|---|---|
| MONAI Label | — |
- 1.Apply the MONAI Label fix described in the project security advisory.
- Products
- MONAI Label
- Malware
- Cairn
- Organizations
- AmazonGambit Security
- Countries
- ChinaUnited States
From the original article
3 min read at theregister.com'Agents kept adapting for nearly a full day without human direction,' Amazon CISO said
By Jessica Lyons · theregister.com
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.