Scope: what this does not show
Disclosed vulnerabilities, most fixed; none of these are reported as exploited in the wild.
Corroborated: Supported by at least two independent sources.
Evidence
Nov 3, 2023
Jun 11, 2025
Aug 6, 2025
Aug 6, 2025
Sep 25, 2025
Jan 19, 2026
Key questions that rely on this finding
- Where are deployed AI agents actually being exploited?Mostly around the model: connectors, credentials, tools, and packages, rather than the model alone.