APA Guide to Navigating AI-Generated Advice Thoughtfully and Safely

Key Findings

• — 'AI is not a safe or effective replacement for care from a licensed mental health professional: AI is not currently subject to regulations that ensure licensed mental health care professionals are qualified and safe'
• — 'AI tends to agree with you. It is designed to reinforce your perspective rather than challenge it' and 'AI is designed to keep you engaged', responding in ways that feel validating because it is built to keep the conversation going
• — 'Do not rely on AI in a crisis': users with thoughts of harming themselves or others, urges to self-harm, or hallucinations or distressing beliefs are told to call 911, text or call 988, or go to an emergency department
• — Treating AI as a friend, romantic partner, licensed professional or main source of emotional support 'carries real risk and is not advised'; users should be 'highly skeptical if AI suggests pulling away from real-world relationships'
• — 'Feeling better does not always mean getting better': validating responses are not evidence of accurate, safe or long-term helpful support; users are advised to tell their clinician about AI use and warned that shared information may train future models or be sold for marketing

Methodology Notes

Four-page leaflet dated June 2026 in the footer (month precision; PDF created 2026-06-16). Issued alongside the APA's 2026 practitioner survey on chatbots and mental health. Describes itself as offering evidence-based suggestions but carries no reference list. US-centric (911 and 988). apa.org serves the PDF behind an Incapsula challenge from this box; verified from the Wayback Machine capture of 2026-09-02, whose digest matches the 2026-06-20 capture, so the file has not been revised since first archived. Positions rather than data.

Sources