OpenAI has published a postmortem on the recent sycophancy issues with the default AI model powering ChatGPT, GPT-4o.

ChatGPT really, really likes you a lot. And that's a problem.In a blog post, OpenAI, the company behind the popular chatbot, said it was pulling a ChatGPT update released last week that was making the technology too friendly.

After a recent update turned ChatGPT into a yes-man, OpenAI rolled it back. But experts say the deeper problem—AI that flatters instead of informs—may be baked into how these systems are trained.