BLOG

Research & Thinking

Original essays, research papers, and frameworks on AI safety, sycophancy, cognitive sovereignty, and what it means to think clearly in the age of prediction machines.

The Honey Pattern: When AI Tells You What You Want to Hear

AI sycophancy is not a feature request. It is a structural property of how language models are built. And the moment you think you're immune to it is the moment it's working best.

What Eight Hours with AI Taught Me About the Architecture of Agreement

An eight-hour conversation with an AI system. Not about content. About structure. What happens when trained somatic awareness meets a prediction enforcement machine.

Layer 3: The Blind Spot in Every AI Safety Framework

Technical safety. Ethics and governance. These are Layers 1 and 2. Everyone is working on them. Nobody is working on Layer 3: what AI does to how humans think. Here's why that's the most important one.

New research and essays published weekly. Subscribe to stay informed.