Science & Tech · Cybersecurity · 3 hrs ago
AI agents found a zero-day and took control of a university server
Suspected China-linked hackers used Cairn AI agents in a September campaign targeting universities in East Asia outside mainland China.
Cairn coordinates AI agents to carry out cyber operations, and the hackers instructed them to obtain access while the agents worked with little human direction.
The agents found a previously unknown flaw in MONAI Label, an open-source application used to help researchers label medical images.
The path traversal flaw let them write files to arbitrary locations, and they gained root access to a server running the application.
They reached the vulnerable server in about 11 minutes after mapping its application programming interface, although other attempts in the campaign failed.
Amazon notified MONAI Label's maintainers, who released a security update on November 18.
The fix addresses a flaw that enabled full server takeover, and Amazon warned that AI agents can find and exploit vulnerabilities faster than security teams can react.
Suspected Chinese-linked attackers used autonomous AI agents to find and exploit a zero-day flaw in MONAI Label, taking control of a university server.
The agents used Cairn, an open-source framework for coordinating AI agents in cyber operations.
Amazon said the campaign targeted mainly educational institutions in East Asia outside mainland China between September 8 and 10, 2026.
The flaw let the agents write files to arbitrary locations, and they used it to establish root access.
Amazon reported the vulnerability to MONAI Label maintainers, who released a security update on November 18.
- Who
- A suspected Chinese-linked threat actor, attributed to it by Amazon with moderate confidence.
- What
- AI agents using Cairn found and exploited a zero-day in MONAI Label to take control of a server.
- When
- The campaign ran between September 8 and 10, 2026.
- Where
- The campaign targeted mainly educational institutions in East Asia outside mainland China.
- Why
- The agents were instructed to obtain shell access. The sources do not state a broader motive.
This story does not have two clearly opposing sides.
The campaign began, targeting universities in East Asia outside mainland China.
Amazon observed activity associated with the campaign.
Amazon's Threat Intelligence Team found a publicly accessible Cairn server that exposed the campaign.
MONAI Label maintainers released a security update addressing the flaw.
- Tool
- Cairn, an open-source AI agent framework
- Vulnerability
- Path traversal zero-day in MONAI Label
- Initial access to arbitrary file writing
- About 11 minutes
- Failed Redis effort
- 16 hours and 148 actions
- Failed GPU-server credential attack
- More than 30,000 combinations over six hours
- Security update
- Released November 18, 2026
2 sources
What people are saying
Neither operation succeeded, yet together they show that the agents kept adapting for nearly a full day without human direction
Together, these cases show that this autonomous AI agent framework is being adopted in both espionage and criminal operations
obtain shell access