top of page

The First AI SOC Summit: What Happened When Practitioners Get Together

  • Writer: Mike Dupuis
    Mike Dupuis
  • Mar 6
  • 2 min read

Updated: Mar 6

Monzy Merza at AI SOC Summit | March 3 2026
Monzy Merza at AI SOC Summit, March 3 2026

On March 3, security practitioners walked into the Hyatt Regency Tysons Corner and spent a full day being honest with each other about AI in the SOC. No vendor pitches. No AI theory. Just people who actually do this work, sharing what's working, what's failing, and what's genuinely dangerous.


The keynote is below. Watch it first, then I'll tell you what else happened in the room.




What Else Happened

The breakout sessions and hackathon were where the day got its hands dirty.


  • Gary Lopez (Tinycode) showed live attacks against AI systems, hidden prompt injections, agent goal manipulation, AI-orchestrated exploitation, and handed the room the tools to test their own. 

  • Lipyeow Lim wrapped the morning session with a session dedicated to the changing security landscape complete with a Crogl demo, no curated data, no planted questions, setting up an afternoon for practitioners stress-testing AI claims. That felt right.

  • Ed Albanese (ThirdLaw) with the question organizations are starting to face: what does it mean to monitor AI behavior, not just security events? 

  • Brennan Lodge (BLodgic) tackled the GRC-to-SOC gap with a RAG-based approach that was specific and implementable. 

  • Eric Zietlow (DeepTempo) made the case for behavioral timeline models where signatures and anomaly detection both fall short. 

  • Raja Sekhar Rao Dheekonda (Dreadnode) dropped a number that stopped people mid-conversation: 186 jailbreaks in 137 minutes. His point wasn't to scare anyone. It was a forcing function for how seriously we need to take AI red teaming at scale. 

  • And Andrew Heibel closed the breakout track while going deep on LLM agents operating continuously in complex environments, practical, not theoretical.


The hackathon ran in parallel giving attendees the opportunity to extend a secure AI platform into the tools they use on a daily basis and build the skills they need to work at agentic speed. People weren't watching. They were building.



Thank You


To our speakers: you made the day what it was. The honesty you brought to the room is exactly what this community needs more of.


To our sponsors, CMAI, Cribl, ReversingLabs, Splunk, and ThunderCat Technology, thank you for backing a room full of practitioners who came to work, not watch.


The conversation is continuing in the Crogl Community on Slack. Come join us.

Comments

bottom of page