AI in 2025: A Look Back With Allen Frances, MD

Allen Frances, MD, discussed the rapid expansion of artificial intelligence (AI) and its implications for psychiatry in 2025. He described recent advances in chatbots as technically impressive but clinically hazardous, citing growing evidence that AI interactions had exacerbated psychosis, suicidality, eating disorders, and conspiracy thinking, and had fostered behavioral addiction. Media attention and litigation related to AI-associated harms had increased public awareness and had begun to pressure AI companies to improve safeguards, although Frances emphasized that current safety standards remained grossly inadequate.¹

Frances argued that clinically informed stress testing of chatbots should have preceded their widespread release and was now urgently required, given that more than a billion individuals were already using these tools. He noted that patients, particularly younger individuals and increasingly older adults, were using chatbots as therapists or companions because of accessibility, convenience, and cost. He maintained that mental health professionals could not resist this trend and instead needed to adopt hybrid models that leveraged chatbot strengths while preserving uniquely human clinical skills.

Clinically, Frances advised psychiatrists to routinely assess chatbot use during evaluations and to consider chatbot-induced psychopathology in the differential diagnosis. He suggested that supervised use of chatbots for between-session support and therapeutic assignments could be beneficial if transparently integrated into care.

Frances opposed development of a DSM-6, warning that diagnostic expansion had fueled overdiagnosis and diverted scarce resources from the severely mentally ill. He called for stronger regulation of AI, especially to protect children and privacy, and criticized the profit-driven ethics of AI companies. Finally, he expressed skepticism about biologically focused psychiatric research, advocating instead for investment in clinical, psychotherapy, and services research, while cautioning that unregulated AI development posed profound societal and existential risks.

References

1. Ummer-Hashim S. AI chatbot lawsuits and teen mental health. American Bar Association. October 27, 2025. Accessed December 22, 2025. https://www.americanbar.org/groups/health_law/news/2025/ai-chatbot-lawsuits-teen-mental-health/

2. Guarino M. AI chatbot dangers: are there enough guardrails to protext children and other vulnerable people? ABC News. November 2, 2025. Accessed December 22, 2025. https://abcnews.go.com/Technology/chatbot-dangers-guardrails-protect-children-vulnerable-people/story?id=127099944