Cleandra Waldron, a counselling psychologist, shares the troubling patterns emerging with clients in her therapy room. Clients increasingly reveal details of their conversations with LLMs as they turn to AI for mental health support. Often unaware of the risks to privacy and dependence, they reveal intimate details of their lives and even medical data. A recent Sky News article reported that an alarming 1.2 million people had discussed suicide with ChatGPT. The ease of 24/7 support without wait times during an unprecedented mental health crisis—which largely operates in a regulatory void—has dangerous implications for user safety. Human psychological services have taken years to build safeguards and protections that clients take for granted, while AI poses as a therapist without any of the regulatory safeguards and protective guardrails that a human therapeutic relationship is bound by.
This article examines the real-world implications of AI therapy through the lens of clinical practice, revealing alarming gaps in data privacy, the dangers of AI “hallucinations” in therapeutic contexts, and the fundamental tension between business models optimized for engagement and the wellbeing of users. Drawing on recent legal actions against big tech AI companies, emerging research, and first-hand accounts from therapy sessions, it carefully asks critical questions: What do we lose when algorithms replace human connections? How do we balance the increased demand for cost-effective therapeutic services with common-sense protections that keep users safe?
The future of AI and its implications for therapy remain unclear. This article poses more questions than answers but aims to increase awareness and promote further research in the field of AI therapy, encouraging the implementation of common-sense policies that protect users from a therapist that never gets sick or goes on holiday.