Overall, OpenAI found that "mental health conversations that trigger safety concerns, like psychosis, mania, or suicidal thinking, are extremely rare." But because ChatGPT’s user base is so vast, even ...