Most AI safety conversation focuses on what's happening inside the model. In this episode, Dr. Ilan Strauss, co-director of the AI Disclosures Project, shares why he thinks that's the wrong place to look. The bigger risks, he argues, live in the market — in how AI products get optimized for engagement, how infrastructure concentration replicates the platform economy's worst outcomes, and why content creators are unlikely to be compensated without new market mechanisms. If the web taught us anything, it's that open infrastructure and closed infrastructure produce fundamentally different power distributions. AI is at that same fork.