The rogue agent problem: When AI adoption outpaces your security
Presented by: Milo Gilad
Originally aired on October 24 @ 6:00 AM - 6:30 AM EDT
AI adoption doesn't roll out — it spreads. One approved tool becomes ten unapproved ones. One experiment becomes a production deployment nobody secured. For security investigation and response teams, every ungoverned AI deployment is a blind spot — new attack surface you can't monitor, new data flows you can't inspect, and new behaviors your detections weren't built to catch. This session covers the real patterns behind AI sprawl — shadow AI, ungoverned deployments, supply chain risk, and data leakage — through the lens of security investigations and incident response. We'll share the frameworks Cloudflare's security response team built to identify, investigate, and respond to AI-driven risk at scale.
English
