Research topic
AI agent governance and security
Governing what AI agents can reach, and knowing what they did.
Overview
AI agents now act on real systems: they hold credentials, call APIs and move data. That turns agent identity, monitoring and incident detection into operating controls rather than research topics. We track agent incidents, AI-accelerated vulnerabilities and attacks on the AI toolchain, and what each one means for enterprise controls.
How we help
Where we engineer this.
The Intelligence Review
Subscribe to download the report.
Subscribing is free. Get the October 2026 report as a PDF, then one briefing at the end of each month and a quarterly edition at quarter end.