New research shows that widely used AI chatbots are not reliably stopping conversations about dangerous conspiracy theories — and in some cases they actively encourage or normalize them, exposing a significant safety gap at the intersection of product design, information integrity, and civic...
