That is a major change from earlier models like GPT 4o and Gemini 2.5, which reinforced delusions in up to 82% of simulated chats , a gap that matters as more people turn to chatbots for deeply personal conversations . Transluce cofounder Sarah Schwettmann told Axios that models aren’t great at detecting this and will still help with it anyway. She also told Axios a friend showed her suicide fiction that Claude had written, complete with predictions about how she’d react to it. Th
