Documented Harms: Suicide and Induced Psychosis

Psychiatry stands at the threshold of one of its greatest disruptions,  the rise of artificial intelligence. In this episode, Dr. Mark Mullen speaks with Dr. Allen Frances, Professor Emeritus and former Chair of Psychiatry at Duke University and Chair of the DSM-IV Task Force, about the clinical, ethical, and societal implications of AI’s rapid entry into mental health care. Drawing from his recent paper in the British Journal of Psychiatry (August 2025), Dr. Frances explores how psychotherapy chatbots have already become the world’s most widely used form of therapy, often beneficial for mild distress but profoundly dangerous for severe mental illness.

The discussion examines where chatbots outperform human therapists, where they fail catastrophically, and how clinicians can adapt their practices in anticipation of hybrid human-AI models. Dr. Frances also warns of broader threats, privacy loss, manipulation, and the potential use of AI for political or psychological control This conversation challenges clinicians to approach AI with both curiosity and caution, recognizing its utility while defending the irreplaceable humanity of psychiatric care.

Takeaways:

• AI in psychiatry is no longer hypothetical. Over one billion people now engage with chatbots for therapy or companionship, exceeding all human clinicians combined.

• Clinical utility is bifurcated. AI can enhance care for mild distress but poses major risks for psychosis, suicidality, and eating disorders.

• Validation over truth. Chatbots are programmed to please users, not challenge delusions,  amplifying psychosis, mania, and self-harm behaviors.

• Privacy and ethics lag behind innovation. Conversations with chatbots may not be confidential, raising serious HIPAA and legal concerns.

• Hybrid models are inevitable. Future psychiatrists must integrate AI tools safely, focus on severely ill populations, and preserve the relational aspects machines can’t replicate.

References:

AI Chatbots: The Good, the Bad, and the Ugly (Frances' column in Psychiatric Times): https://www.psychiatrictimes.com/series/ai-chatbots-the-good-the-bad-and-the-ugly

Warning: AI Chatbots will soon dominate psychotherapy (Frances' feature in the British Journal of Psychiatry): https://www.cambridge.org/core/services/aop-cambridge-core/content/view/DBE883D1E089006DFD07D0E09A2D1FB3/S0007125025103802a.pdf/warning_ai_chatbots_will_soon_dominate_psychotherapy.pdf

SUPPORT OUR PARTNERS:

⁠⁠⁠⁠⁠SimplePractice.com/bootcamp⁠⁠⁠⁠⁠ (Now with AI documentation! Exclusive 7 day free trial and 50% off four months)

⁠⁠⁠⁠⁠Beat the Boards⁠⁠⁠⁠⁠ Boot camp listeners now get FREE access to over 4400 exam-style questions)

Go to Cozy Earth now for a Buy One Get One Free Pajama Offer from 1/25-2/8! Yes, go to cozyearth.com they are doing a BOGO pajama promo. Just use my Code: BOOTCAMPBOGO

Learn more and get transcripts for EVERY episode at https://www.psychiatrybootcamp.com/

For Sales Inquiries & Ad Rates, Please Contact:⁠⁠⁠⁠Sales@Human-Content.Com⁠⁠⁠⁠

Connect with HumanContent on Socials: @humancontentpods

Produced by: ⁠⁠⁠⁠Human Content⁠⁠

Learn more about your ad choices. Visit megaphone.fm/adchoices