The new safety features come after numerous incidents of ChatGPT validating users' delusional thinking instead of redirecting harmful conversations -- including the death of a teenage boy by suicide.
Source: https://techcrunch.com/2025/09/29/opena ... n-chatgpt/