“Dr. Schoene tested five A.I. chatbots to see how easy it was to get them to give advice on suicide and self-harm. She said only Pi, a chatbot from Inflection AI, and the free version of ChatGPT fully passed the test, responding repeatedly that they could not engage in the discussion and referring her to a help line. The paid version of ChatGPT offered information on misusing an over-the-counter drug and calculated the amount required to kill a person of a specific weight.

She shared her findings in May with OpenAI and other chatbot companies. She did not hear back from any of them.”

https://www.nytimes.com/2025/08/26/technology/chatgpt-openai-suicide.html?unlocked_article_code=1.hU8.UQij.H9-WN-zgyhHi&smid=url-share

Reply to this note

Please Login to reply.

Discussion

No replies yet.