AI, LLM, and chatbot security
Untrusted documents and webpages can steer an AI assistant away from its task. Learn how prompt injection works and how to limit its impact.
Read the articleInsights
How the attack actually works.
Source-backed writing on attacker behavior, authorization, and the decisions that change an outcome.
Counterintelligence in public view
MGM lost roughly $100 million to a technique that predates the computer. The recruitment cycle behind it is documented, teachable, and almost entirely absent from security awareness training.
Read the articleThe same material, taught in depth.
Every post here connects to a module in the Tradecraft Series, delivered with demonstrations, decision exercises, and completion evidence.
Prepare. Anticipate. Protect.