AI agent exploits medical imaging zero-day in 11 minutes
November 22, 2026
Amazon Threat Intelligence identified a Chinese-nexus actor using autonomous agents to find and exploit zero-day vulnerabilities. One agent gained shell access to a medical-imaging application within 11 minutes using a single command.
HOW THIS AFFECTS YOU
●
builderYou need to sandbox any agentic workflows to prevent them from accessing system shells.
●
policyRegulators will likely target unsecured agentic tool use as model capabilities increase.