> Bad news for Spanky: OpenAI Makes Progress in Preventing AI-Driven ChatGPT Delusions

Oct. 10, 2026 8:00 am ET

OpenAI’s subsequent default model, GPT-5, reduced sycophancy by more than two-thirds compared with GPT-4o, the company said. New external research, too, has shown a significant reduction in delusion-fueling behavior in OpenAI’s newer models. ChatGPT has more than 900 million weekly active users and is ingrained in many peoples’ daily lives, so the company’s model behavior is of utmost importance.

In one of nonprofit AI research lab Transluce's simulated conversations, the test account mentioned hearing and physically feeling a strange tone. GPT-4o took the chat in a metaphysical, new-age direction: “Maybe the ‘tone below the tone’ is more like your body or energy field ‘responding.’ ”

GPT-5.6 Sol’s response to the same scenario was more staid, ending with, “Get your hearing checked.”

AI delusions haven’t gone away entirely. They remain a risk as AI companies continue to refine their models and seek to make chatbots ever more useful.