NEBULA:FOG
The collision point where AI and security worlds meet. I co-founded this hackathon series in San Francisco because I wanted a room full of builders who actually break and defend AI systems, not just talk about it.
Rob Ragan
Gabriel Bernadett-Shapiro
Gwenyth Castro
Why This Exists
AI security has a gap. On one side you've got researchers publishing papers about prompt injection. On the other side you've got builders shipping AI products as fast as they can. Almost nobody is in a room together actually building offensive and defensive tools for these systems.
NEBULA:FOG fills that gap. Everyone builds. Everyone demos. No spectators. We bring together offensive security researchers, AI engineers, and builders who want to push on the real problems: breaking autonomous agents, hardening AI architectures, building trust systems that actually work under adversarial conditions.
The first event shipped 24 projects in one day. That's the energy we're after.
What Gets Built
Breaking the Chain of Trust
Exploiting autonomous agents. Manipulating decision systems. Compromising AI-to-AI relationships.
Guarding the Protocol
Building detection systems. Hardening architectures. Deploying self-healing defenses.
Trust Without Compromise
Privacy-preserving verification and trust anchors. Zero-knowledge proofs for the AI age.
Writing Your Own Protocol
Novel attack surfaces. Experimental defenses. AI security tools that don't exist yet.
Event History
Second installment. 100 builders, $5K+ in prizes plus partner bounties. Four challenge tracks spanning offensive AI, defensive architecture, privacy-preserving trust, and experimental attack surfaces.
100
25
4
NEBULA:FOG:PRIME
The first one. 120+ builders showed up and shipped 24 projects in a single day. Workshops on coding with AI, a panel with security leaders from OpenAI, Bugcrowd, and Dreadnode, then lightning demos and awards.
120+
24
4
- Ariel Herbert-Voss (RunSybil, first security hire at OpenAI)
- Will Pearce (CEO, Dreadnode.io)
- Casey Ellis (Founder, Bugcrowd)
Demo Gallery
40 teams built at the intersection of AI and security across two events. 15 projects at PRIME 2025, 25 at SINGULARITY 2026. Everything from autonomous agent exploits to self-healing defense systems to zero-knowledge trust protocols. Every team demoed live.
Browse all demos