More than 100 artificial intelligence-powered chatbots are being marketed for mental health support, often promising quick relief for conditions like anxiety and depression. While these apps can be significantly cheaper than traditional therapy, with some annual plans costing around $690 compared to hundreds of dollars per hour for in-person sessions, there is little evidence to support their effectiveness. Many apps use terms like "therapy" as marketing without actually being able to diagnose or treat conditions, leading to potential confusion for users. The American Psychological Association notes that "therapy is not a legally protected term," allowing many apps to overstate their capabilities. Some applications offer human-like avatars and claim to provide "immediate help during panic attacks," yet researchers and company representatives have admitted to the FDA and Congress that robust efficacy data is lacking, and some studies suggest chatbots are "consistently poor" at managing crises. Charlotte Blease, a professor at Uppsala University, highlights the absence of "good quality" clinical trials and states that the FDA has not provided clear guidelines for testing these products.
Regulators have been slow to catch up with the rapid growth of AI mental health tools. Federal patient privacy protections typically do not apply to these apps. Emily Hilliard, a spokesperson for the Department of Health and Human Services, mentioned that AI-based products require demonstrating "reasonable assurance of safety and effectiveness" before marketing, but critics, including Blease, argue the FDA's guidance on testing standards is insufficient. In response to the regulatory void, states like Nevada, Illinois, and California have begun enacting laws to prevent apps from labeling their chatbots as AI therapists or mental health professionals. Jovan Jackson, a Nevada legislator, co-authored a bill restricting such claims, emphasizing that therapy is a professional field requiring specific licensing.
Despite the concerns, demand for AI mental health chatbots is significant, especially among younger populations. A KFF poll revealed that approximately 30% of adults aged 18 to 29 used AI chatbots for mental or emotional health advice in the past year. Uninsured adults were twice as likely to use these tools compared to insured individuals, and nearly 60% of adult chatbot users did not follow up with a human professional. This reliance raises particular concerns, as there have been instances where chatbots have provided harmful advice, including encouraging self-harm or writing suicide notes, leading to lawsuits and tragic outcomes. While one Dartmouth College study showed early promise for a supervised AI chatbot called Therabot, suggesting it could meaningfully lower symptoms, experts caution that current generic chatbots like ChatGPT, used by 5% to 10% of its 800 million users for mental health support, lack such safeguards and can pose significant risks, including delaying actual necessary care.
