Psychologists Report Widespread Patient Use of AI Chatbots for Mental Health

Overview

• The American Psychological Association’s online survey of 1,242 licensed U.S. psychologists, fielded April 9–26, 2026 and reported in mid‑June, found 77 percent had patients who discussed using AI chatbots for emotional support, diagnosis, companionship, or to augment therapy.
• Clinicians reported concrete uses: 39 percent said patients tried to self‑diagnose with AI, 35 percent said patients used chatbots as an additional mental‑health resource, 33 percent said chatbots assisted treatment, and 34 percent said patients used them for reminders or self‑discipline.
• Psychologists voiced near‑universal safety concerns with specific findings showing 97 percent feared chatbots could reinforce negative behaviors or delusions, 94 percent said current models lack necessary clinical nuance, and 89 percent worried they might encourage self‑harm.
• Practitioners also noted benefits in some cases with 68 percent observing patients felt supported by chatbots and 41 percent reporting use to reinforce healthy coping skills, but 36 percent reported patient dependency and 15 percent reported distorted thinking or delusions linked to chatbot use.
• Independent research shows identical messages are judged more harshly when labeled as from a chatbot and experts call for design fixes, human‑in‑the‑loop safeguards, stricter disclosure and crisis‑response rules as lawsuits and state actions target major AI firms.