I'm Steve Brodson. I'm a cybersecurity architect focused on AI safety and security. I experiment with AI systems, consult with organizations navigating AI risk, and teach practitioners how to think clearly about threats that don't fit traditional security frameworks. I write a bi-weekly newsletter called AdversarialMinds.
The AI security conversation has two loud camps. One warns AI will destroy humanity. The other insists it will save us. Both are arguing about systems that don't exist yet. Meanwhile, real problems are compounding in production: shadow AI, prompt injection, credential exposure, agentic systems whose emergent behavior nobody can predict. That's the gap I am trying to close.
What I actually do
Consulting. Embedded work with security, engineering, and leadership teams. Mapping the AI already running in their environment — including the tools IT never approved and the AI features quietly embedded in everyday applications. Threat-modeling the non-deterministic surface. Building governance that survives contact with production.
Speaking. Keynotes and workshops for boards, industry summits, and internal all-hands. No jargon, no doom — just a clear picture of what's happening, what's knowable, and what to decide.
Writing. The AdversarialMinds newsletter is where I think out loud. Bi-weekly analysis of what's actually happening in AI security, written for people who are responsible for it. Plain language, honest uncertainty, practical steps.
Tooling. Small, practical tools that help teams see and govern the AI already running in their environment. Released quietly, maintained honestly.
The position I write from
AI security is a discipline of managing real exposures realistically. That means applying the standards we already have, being comprehensive about where AI actually lives in an environment, and finding new ways to manage the novel risks that keep compounding. It's not a departure from traditional security — it's traditional security done honestly against a faster-moving surface.
Anyone who tells you they have clean answers to the hard questions in AI security is either not being honest or not being careful.
How to reach me
Email: steve.brodson@blocksentience.com
LinkedIn: in/stevebrodson
X: @adversarialmnds
Substack: @adversarialminds