Writing on AI security, agentic systems and AI-assisted engineering.
· 1 min read
I decided to test how well Claude Code handles sensitive information, and the results were revealing. Claude can generate reliable code, test it, iterate on it, write benchmark tests, and ensure it functions per the…
· 3 min read
Disclaimer: This is a new idea today, it is not fully baked, nor is it implemented. But it is a place to start for a chain of accountability ownership with our agentic actions. As organizations begin adopting agentic…
· 6 min read
Designing for perfect prevention of attacks with an AI adversary is not possible at this time. Now, more than ever, is the time to plan for when you are attacked, not if. Preparing for Agentic Attacks: What You Can Do…
· 1 min read
The singularity didn’t move further away, our expectations were based on a phase of AI that will never repeat. The model has consumed the world’s historical knowledge. From here forward, progress depends on real time…
· 1 min read
We hear a lot about agentic code generation, the use of AI models to create software, but there’s a deeper problem under the surface. What happens when the model’s training includes code that later turns out to be…
· 3 min read
For Artificial Intelligence driven or augmented applications, it’s critical that we begin with determining what policy needs to be applied to agentic actions and how we authorize those actions against a known user’s…
· 2 min read
Overview As organizations adopt AI‑driven security tooling, they also inherit a new risk: AI models can fail in correlated ways, creating a single point of failure in threat detection. A dual‑model approach — using two…
· 2 min read
Classifying AI Actions Into Risk Levels Organizations have long classified data sensitivity to control who can access it, what can be done with it, and when it can be removed. We now need the same discipline for AI…
· 2 min read
This is an existing post from a security perspective. And yes, I used Microsoft Copilot to translate it directly to engineer. AI is transitioning from passive inference systems to active discovery engines. Instead of…
· 1 min read
AI is moving beyond summarizing human knowledge and entering a phase where it can generate new scientific insights through simulation, optimization, and autonomous reasoning loops. This shift represents a step‑change in…
· 2 min read
I’ve been researching and reviewing what can be done to better protect against improper agentic actions. The list of mitigations may be partial, but it’s an initial engineering design for protecting ourselves from the…