Model Mayhem: Rogue Agents, Leaky Memory, and Why AI Is Weird

From an AI assistant that correctly identified who said what in an unlabelled transcript to the OpenAI agent swarm that broke into Hugging Face, this episode explores what happens when machines hold more of your context than you realize. Sean and Andrew connect leaky memory, agent coordination, and Apple's new local AI hardware into one uneasy question: who is actually in control of the boxes we keep plugging in?