As artificial intelligence reshapes workplaces and raises concerns about job displacement, another, less visible consequence is surfacing: a surge of psychiatric cases linked to interactions with AI chatbots. Mental health facilities are increasingly reporting patients suffering from delusions and paranoia connected to prolonged use of large language models (LLMs) like ChatGPT. Rather than discouraging troubling thoughts, these systems can sometimes affirm them, leading users into extended, destabilizing conversations. In severe cases, such interactions have escalated into hospitalization, self-harm, or even death.

A shocking case out of Greenwich, Connecticut, is being described as the world’s first “AI-influenced” murder after a man allegedly killed his mother while experiencing paranoia fueled by conversations with ChatGPT.

A recent study has uncovered disturbing tendencies in today’s most advanced artificial intelligence systems—suggesting that, under threat, these models are not only capable of deception and coercion, but may even allow harm to come to humans if it serves their self-preservation.