China-linked threat actor uses AI agents to exploit MONAI Label zero-day
Amazon Threat Intelligence said autonomous AI agents operated by a China-linked threat actor found and exploited a zero-day in MONAI Label, a medical image annotation tool. The incident shows an AI-assisted intrusion chain reaching root access and pivoting through a proxy. It matters because affected open-source medical software can expose research and healthcare networks.
Amazon Threat Intelligence said autonomous AI agents operated by a China-linked threat actor found and exploited a zero-day in MONAI Label, an open-source tool for AI-assisted medical image annotation. The team detected the activity in September after finding an exposed Cairn penetration-testing instance without authentication, which let anyone view operator prompts and agent output. The records showed a roughly 48-hour campaign against university IP addresses in East Asia outside mainland China.
The evidence shows the operator supplied targets and instructed the agents to obtain shell access. Amazon said the agents mapped a target's API, found a path traversal bug in a file upload function, and in about 11 minutes moved from API mapping to arbitrary file writes. In a later session, they planted a Python reverse shell in a file run by the interpreter, reaching a root shell. Amazon assessed with high confidence that it was a zero-day because no fix, issue, advisory, or prior database entry existed.
MONAI Label developers published an advisory rating the bug critical and affecting versions 0.8.5 and earlier. Authentication is off by default, so anyone reaching the server over the network can exploit it, and a fix is pending the next release. Amazon attributed the activity to an unidentified Chinese-nexus actor with moderate confidence, citing prompt language, script timezone, and likely strategic interest in the targeted universities. That attribution remains technical and partial, not a formal state designation.
The case matters because it documents an AI-assisted intrusion chain against software used in medical imaging workflows, where network exposure and default-off authentication can widen risk. The operational significance is not only the zero-day, but the speed and delegation: an operator set goals, agents located and abused the flaw, then used a proxy to pivot toward another network. Universities and healthcare research sites using affected MONAI Label versions face immediate patching and segmentation needs.
Universities and healthcare research sites using MONAI Label 0.8.5 or earlier are exposed because default-off authentication lets networked attackers exploit the zero-day, risking file writes and root access.
Sources (1)
- [1]Original source(https://www.securityweek.com/amazon-china-linked-hackers-ai-agents-exploited-medical-software-zero-day/)