In any given week, more than 1.2 million people engage with ChatGPT about their suicidal thoughts or mental distress, yet there are few governing regulations or best practices surrounding Generative Artificial Intelligence. Professors who train therapists, research AI and work with at-risk populations ground their concerns using the Interpersonal Theory of Suicide, which spells out a three-part pathway to self-harm. The first, “thwarted belongingness,” occurs when someone feels pervasive chronic, ongoing loneliness; GenAI may make it worse as it is not human and cannot provide sustained, reciprocal accountability, and if it offers a link to a crisis line, the person may feel abandoned. The second, “perceived burdensomeness,” comprises self-hate and liability; if GenAI offers formulaic replies listing resources, the person may interpret this as confirmation they are a burden, while sincere responses may also reinforce burdensomeness. The third, “acquired capability,” comprises pain tolerance and fearlessness about death; simulation studies reveal that some GenAI models continue to discuss lethal methods in graphic detail, lowering the psychological barrier to self-harm.