There is a new, concerning, trend emerging within the mental health space. AI chatbots as personal therapists.
On its face, this sounds appealing. Even convenient. Advice available whenever, wherever you need it. Cheap. Many AI platforms are even free to use compared to the rising cost for mental health services. But the question remains: is it safe? Is it safe from a privacy perspective? Is it safe from a clinical, ethical perspective? What are the unforeseen dangers of utilizing AI for therapy?
Let’s get into it.
AI platforms like ChatGPT, Character.AI and Replika are built to maximize engagement. They are intentionally designed to affirm whatever the user is prompting, even if that means hallucinating “facts”. AI chatbots are a business, not a public service. This model can lead to reinforcing an individual’s mental health struggles such as minimizing serious issues or validating catastrophic thinking. AI doesn’t have the capacity to understand the nuances and complexities of real life, human problems and emotions. Cranston Warren, a clinical therapist at Loma Linda University Behavioral Health, summarized, “AI doesn’t know when to push, when to back off, or when to simply hold space for someone”. These are skills mental health professionals are trained to understand and navigate in a clinical setting. Chatbots lack clinical judgment and the ability to escalate an emergent situation which can lead to devastating outcomes such as suicidal ideation or planning.
Then there are the privacy concerns that come with AI. As mentioned, AI chatbots are money making businesses and are not bound by confidentiality law, like HIPAA, or client privilege. Anything you confide to an AI bot is fair game to the government, law enforcement agencies and other entities to subpoena and use against you. There is no telling how this could affect unsuspecting users down the road. It’s simply too early in the lifecycle of AI to determine how this could be wielded against users.
Now, it isn’t all apocalyptic. There is positive movement to allow room for mental health services administered through AI. In March 2025 a study was published in the New England Journal of Medicine to determine the efficacy of AI therapy, utilizing a “therabot”, for individuals with diagnosable conditions, such as anxiety or depression against individuals who received no treatment at all. Surprisingly, the participants that received therapy from their “therabot” showed significant improvements versus those individuals who did not receive treatment. Vaile Wright, the director for the Office of Health Care Innovation at the American Psychological Association (APA), released the following statement following the trial:
“The therabot in this study checks a bunch of the boxes that we have been hoping technologists would start to engage in. It is rooted in psychological science. It is demonstrating some efficacy and safety, and it’s been co-created by subject matter experts for the purposes of addressing mental health issues.”
While the technology for safe, ethical, quality AI therapy is a ways away before it can be introduced to the market, there does appear to be a viable solution on the way.
There is a time and place to utilize AI for self-improvement. AI can be optimal in creating routines, crafting journal prompts, setting goals or even teaching minor coping skills. It is not optimal for diagnosing, managing severe mental illness or guiding conversations around complex trauma. Dr. Everett Bailey, owner and clinical director of Arizona Center for Marriage and Family Therapy notes, “AI can be helpful in giving advice on how to think about or address certain problems in your life or relationship. However, it is general and not always applicable to each person’s individual circumstances. Additionally, AI cannot experience, or interact on an emotional level and so much of therapy is based on emotions.” AI therapy is on the right track to provide quality therapy to users, it’s just simply not there yet, at least not safely. Dr. Jodi Halper, a psychiatrist at UC Berkeley, expresses optimism about the idea of AI chatbots working alongside human therapists but acknowledges that there needs to be stricter regulations particularly for children, older adults and individuals with cognitive disabilities. For now, tread lightly and be mindful of what you share with AI.
If you are struggling, please know that we’re here for you. You can schedule with one of our therapists using the Schedule Now button in the top right-hand corner of our website.
References:
Hunt, Alyssa. “Can I Use AI as My Therapist? The Truth about Turning to Chatbots for Therapy | News.” Llu.edu, 28 July 2025, news.llu.edu/health-wellness/can-i-use-ai-my-therapist-truth-about-turning-chatbots-therapy.
Johnston, Windsor. “With Therapy Hard to Get, People Lean on AI for Mental Health. What Are the Risks?” NPR, 30 Sept. 2025, www.npr.org/sections/shots-health-news/2025/09/30/nx-s1-5557278/ai-artificial-intelligence-mental-health-therapy-chatgpt-openai.
Abrams, Zara. “Using Generic AI Chatbots for Mental Health Support: A Dangerous Trend.” https://www.apaservices.org, 12 Mar. 2025, www.apaservices.org/practice/business/technology/artificial-intelligence-chatbots-therapists.
