Technology news and discussion on TalkFuse.
AI agents exploited a MONAI Label flaw to seize a university server
Suspected Chinese-linked attackers used the open-source AI penetration-testing framework Cairn to find and exploit a previously unknown flaw in MONAI Label, gaining control of a university server. The incident shows how AI agents can adapt through complex cyber-attacks with little human direction, raising concerns about their use in espionage and crime.
Amazon said the campaign ran for 48 hours from 8 September and targeted universities in East Asia. Before finding the flaw, agents spent about 16 hours on 148 unsuccessful actions against a Redis server and tried other attacks; a separate operation compiled 80 real user names for credential spraying. Once they identified a path-traversal flaw in MONAI Label, they went from mapping its API to writing files in about 11 minutes, then used it to run a Python reverse shell as root. Amazon notified the project maintainers, who issued a fix and security advisory.
In short
- Cairn agents found and exploited a MONAI Label zero-day.
- They gained root access to a university server.
- Amazon says the agents adapted autonomously for nearly a day.
Read the full article at the source →
Originally published by The Register as “China attackers deployed AI agents that autonomously found and exploited a 0-day to take over a university server”.