The article warns against using ChatGPT for sensitive or high-stakes matters, particularly legal disputes and mental-health crises. Unlike conversations with lawyers, doctors or therapists, chats with AI generally lack legal privilege and may be subpoenaed, exposed or retained, while the system can produce confident but inaccurate advice.

A public database has recorded more than 2,000 court decisions involving fabricated AI citations or arguments, including 1,175 cases involving self-represented litigants and 815 involving lawyers. OpenAI has estimated that 0.15% of weekly users show explicit signs of potential suicide planning—more than one million people—while researchers warn that chatbots can reinforce harmful or delusional beliefs; the company has since revised crisis responses and introduced parental controls.