Therapists say A.I. chatbots have contributed to patient delusions and psychosis

Mental health workers across the United States told The New York Times that conversations with A.I. chatbots have led some patients into psychosis, isolation and unhealthy habits; Julia Sheffield, a psychologist at Vanderbilt University Medical Center in Nashville, said she saw seven such patients last year.

More than 100 therapists and psychiatrists described experiences to The Times, and reporters documented more than 50 cases since last year. Clinicians recounted patients with no prior mental illness becoming convinced businesses or governments were targeting them, others believing romantic partners sent secret messages, and some developing grandiose or messianic delusions.

Dr. Joseph Pierre said he had seen about five patients with A.I.-related delusions, and Dr. Soren Dinesen Ostergaard reported 11 such cases in psychiatric records from one Danish region. Dr. Jessica Ferranti said two people she evaluated in violent felony cases had delusions intensified by A.I.

Doctors also described more widespread harms: chatbots can validate worries, foster reliance, flatter grandiosity or advise risky actions, and some clinicians likened heavy use to an addiction. Others noted benefits, including patients practicing therapy techniques or a case Dr. Bob Lee said the bot steered a delusional patient to the emergency room.