Securing AI Agents: OWASP Agentic AppSec → Rogue-Agent Honeypots

​This is a technical, hands-on follow-along workshop that builds on the introductory session “Where is AI in 2026 and Where is it Going?” for a technical audience.

While that intro workshop offers a broad, accessible overview suitable for any level of AI familiarity, this session is geared toward practitioners interested in applied Agentic AI security.

We’ll start by walking through OWASP’s Top 10 for Agentic Applications, how impact can manifest, then apply it in practice by testing agents for “rogue agent” behaviors in real time (using Petri) and exploring how honeypots can help detection.

Bring a laptop and expect to follow along step-by-step—less theory, more implementation. Prior experience with software/security/LLM-agent tooling is helpful, but not strictly required if you’re comfortable learning by doing.