The Exploding Attack Surface of Agentic AI | F5 and
summary
Claim: moving from single-shot models to tool-enabled, agentic systems makes the attack surface exploded because hundreds of models and agents talk and gain distinct privileges. Enterprises adopt identity governance that treats nonhuman agents like people and apply zero-trust to agent identities, add LLM controls and observability, and run centers of excellence with KPIs to pilot projects. Technical responses include building sovereign data centers (100–1,000+ GPUs), multi-tenant load balancing, and optimizing for first‑time‑to‑token; GPUs enable scale where non‑GPU batch runs do not.
AI-written summary · about
chapters (11)
- 0:00Agentic AI security intro
- 1:02AI performance vs security
- 2:28From single-shot to agents
- 3:52Zero trust for agents
- 5:17Digital sovereignty limits
- 7:04AI governance controls
- 8:49Shadow AI governance
- 10:57AI governance KPIs
- 12:25GPU necessity vs scale
- 13:49Sovereign AI data centers
- 15:11AI innovation rate concerns
related talks
31mAgentic AI Threats: How Multi-Agent Systems Get Exploited
27mFrom Perimeter to Silicon: Why Agentic AI Demands a New Security Model
20mSecurity in the Age of Agentic AI
13mKeynote: Agentic AI Is Running Your Infrastructure
14mAI Sprawl: The Hidden Risk Threatening Your Enterprise Now
39mFrom Demo to Production: Why Agentic AI Systems Fail - and How to Fix Them
30mFuture of AgenticAI: Architecting the Global Control Plane (Keynote)
49mThe Intersection of AI and Cybersecurity