Season 2 · Episode 1959
Why Constrained AI Models Still "Go Rogue"
Your AI assistant promised to only use your documents. Instead, it invented a case law that doesn't exist. Here's why.
My Weird Prompts · Daniel Rosehill
April 3, 202628m 37s
Audio is streamed directly from the publisher (dts.podtrac.com) as published in their RSS feed. Play Podcasts does not host this file. Rights-holders can request removal through the copyright & takedown page.
Show Notes
We all want AI that only knows what we tell it—until it doesn't. In this episode, we explore the technical illusion of "constrained" models and why RAG systems still hallucinate. From financial compliance risks to legal discovery nightmares, discover why your AI's "world knowledge" can overpower your private data and what that means for enterprise deployment.