In short
Direct use of AI specifically for therapy is still small but growing. In a 2026 YouGov survey, 6 percent of US adults said they had used AI for therapy and another 17 percent said they would consider it. Use is higher among younger people: a 2025 study in JAMA Network Open found about 1 in 8 US adolescents and young adults had used AI chatbots for mental-health advice. People most often cite low cost, around-the-clock access, and reduced fear of judgment. Survey data on satisfaction is early and mixed, and most clinicians and the public still rate AI below human therapists for actual therapy.
How many people use AI for therapy
In a 2026 YouGov survey of US adults fielded across late March and early April, 6 percent said they had used AI for therapy, and another 17 percent said they had not but would consider it. Estimates depend heavily on how the question is asked, so it helps to separate narrow therapy use from broader mental-health support. When people are asked specifically about therapy, the numbers stay modest.
Broader mental-health and emotional support is more common than narrow therapy use, and use rises sharply among people who already rely on AI. A survey of 499 US adults fielded in February 2025 by Sentio University researchers, later published in the APA journal Practice Innovations, found that 48.7 percent of respondents who both use AI and report mental-health challenges said they were using large language models for therapeutic support.
These figures are not directly comparable: the YouGov number is the share of all adults, while the Sentio number is the share of a narrower group of existing AI users who also have mental-health challenges. Taken together, they suggest that dedicated AI therapy use across the whole population is still a minority behavior, but that it is meaningfully higher inside the population already turning to AI tools.
Six percent of US adults have tried AI for therapy, yet over a third of people under 30 say they would be comfortable with it. Adoption is running ahead of the evidence, and that gap is exactly why we keep these numbers sourced and current.
Who uses AI for therapy: demographics
Age is the clearest dividing line in the data. A study led by RAND researchers with co-authors from Brown University and Harvard Medical School, published in JAMA Network Open on November 7, 2025, surveyed 1,058 US adolescents and young adults ages 12 to 21 in February and March 2025. The sample included children as young as 12, which is why parents weighing AI therapists for kids and teens should look at this data closely. About 1 in 8 reported using AI chatbots for mental-health advice, and usage was higher among the older end of that range, with roughly 1 in 5 respondents ages 18 to 21 reporting use.
Attitude data points the same direction. In the 2026 YouGov survey, adults under 30 were about twice as likely as older adults to say they would feel comfortable working with an AI therapist, at 37 percent versus 20 percent. That comfort gap helps explain why early adoption skews younger, including toward Gen Z.
Beyond age, reliable demographic breakdowns are limited. Most published surveys focus on age and general attitudes rather than detailed cuts by income, geography, or diagnosis, so claims about exactly which groups use AI therapy most should be treated as provisional.
Why people turn to AI for mental-health support
The reasons people give cluster around access, cost, and comfort. Cost and availability are recurring themes in both the YouGov polling and the Sentio University survey: AI tools are typically cheap or free and available at any hour, which matters when human therapy is expensive or wait times are long. These drivers map onto the broader AI therapy pros and cons: access and comfort on one side, unproven effectiveness on the other.
Reduced fear of judgment is another consistent driver. In YouGov's 2024 polling on AI chatbots and therapy, privacy and anonymity, and the sense that an AI will not judge, came up as reasons people are willing to open up to a chatbot. This lower-stigma framing is part of why some users say they share things with AI that they would hesitate to tell a person.
There are also early signals on perceived helpfulness. In the JAMA Network Open study of adolescents and young adults, among those who used chatbots for mental-health advice, two-thirds engaged at least monthly and 92.7 percent said the advice was somewhat or very helpful.
What surveys say about satisfaction and trust
Satisfaction data is genuinely early and mixed, and the evidence base is limited. On the positive side, the Sentio University survey reported that 63 percent of users said large language models had improved their mental health, and the adolescent study found high self-reported helpfulness among users. The same pattern of high self-reported satisfaction shows up in AI therapist reviews.
Public trust, however, remains cautious. In the 2026 YouGov survey, 55 percent of Americans said AI is worse than human professionals at conducting therapy, 14 percent said it is about equal, and 10 percent said it is better. Looking a decade ahead, half still expected AI to be worse than people at therapy, suggesting skepticism is not only about today's tools. Psychotherapy research gives that instinct some backing: a 2018 meta-analysis of 295 studies with more than 30,000 patients found the alliance between client and therapist to be one of the most consistent predictors of whether therapy works, and that relationship is the ingredient a chatbot can only imitate.
Clinicians report widespread but watchful awareness. An American Psychological Association report published in 2026 found that 77 percent of psychologists had spoken with patients who used AI for support or engagement, and 35 percent reported patients using AI as an additional mental-health professional. At the same time, the large majority of psychologists expressed concerns about patients relying on chatbots.
Concerns and limits in the data
Alongside curiosity there is real worry. YouGov polling from 2024 found that common concerns about AI mental-health tools included data safety (46 percent), lack of access to human support (41 percent), the limited emotional range of chatbots (40 percent), and the risk of misdiagnosis (40 percent). These concerns help explain why interest in trying AI does not always translate into sustained use.
The statistics themselves carry limits. Many figures come from single surveys with modest sample sizes, different definitions of 'therapy' versus 'mental-health support,' and self-selected respondents. Self-reported helpfulness is not the same as a measured clinical outcome, and a 92.7 percent helpful rating from users tells you about satisfaction, not about safety or effectiveness. Controlled AI therapy studies measure outcomes very differently from opt-in surveys.
Because this is a fast-moving area, point-in-time percentages can shift quickly as tools and awareness change. The responsible read of the current data is that AI mental-health use is rising, skews younger, and is driven by access and comfort, while trust and proven effectiveness lag behind the hype.
If you are struggling right now
AI tools are not crisis services and they do not diagnose, treat, or cure mental-health conditions. If you are in crisis or thinking about suicide, call or text 988 in the US to reach the Suicide and Crisis Lifeline, available 24 hours a day.
For ongoing support, AI can be a low-cost, always-available supplement, but it is not a replacement for a licensed clinician. If you want human care, you can browse licensed therapists in our directory and compare options that fit your needs.
Key takeaways
- In a 2026 YouGov survey, 6 percent of US adults said they had used AI for therapy and 17 percent said they would consider it.
- Use skews younger: a 2025 JAMA Network Open study found about 1 in 8 US adolescents and young adults had used AI chatbots for mental-health advice, rising to roughly 1 in 5 among ages 18 to 21.
- Among AI users with mental-health challenges, 48.7 percent reported using LLMs for therapeutic support in a 2025 Sentio University survey published in Practice Innovations.
- People most often turn to AI for low cost, around-the-clock access, and reduced fear of judgment.
- Satisfaction signals are positive but early: 63 percent of users in the Sentio study said LLMs improved their mental health, while 55 percent of Americans in the 2026 YouGov survey still rated AI worse than human therapists.
- The data is limited and emerging, with small samples, varied definitions, and self-reported measures, so figures should be read as trends rather than settled facts.
Looking for care?
Browse licensed therapists in our directory.
Frequently asked questions
How many people use AI for therapy?
It depends on the definition. In a 2026 YouGov survey of US adults, 6 percent said they had used AI specifically for therapy and another 17 percent said they would consider it. Broader mental-health support is more common: a 2025 Sentio University survey found that among AI users who also report mental-health challenges, 48.7 percent used large language models for therapeutic support. Dedicated AI therapy use across the whole population is still a minority behavior, but it is growing.
What percentage of people use AI for therapy?
The most direct national figure comes from a 2026 YouGov survey, in which 6 percent of US adults said they had used AI for therapy. A 2025 study in JAMA Network Open found about 1 in 8 US adolescents and young adults ages 12 to 21 had used AI chatbots for mental-health advice. Percentages vary because surveys define therapy and mental-health support differently.
Who is most likely to use AI for mental health?
Younger people. The 2025 JAMA Network Open study found use was highest among 18 to 21 year-olds, at roughly 1 in 5, and the 2026 YouGov survey found adults under 30 were about twice as likely as older adults to feel comfortable with an AI therapist, 37 percent versus 20 percent. Detailed breakdowns by income or location are limited in the current research.
Why do people use AI for therapy instead of a human?
Survey write-ups consistently point to three reasons: lower cost, around-the-clock availability, and reduced fear of judgment or stigma. Privacy and anonymity also come up as appealing features. These access and comfort factors help explain why interest is highest among younger users, even though most people still rate human therapists higher for actual therapy.
Are people satisfied with AI therapy?
The data is early and mixed. On the positive side, a 2025 Sentio University survey found 63 percent of users said large language models improved their mental health, and a 2025 JAMA Network Open study found 92.7 percent of young users who sought mental-health advice from chatbots said it was somewhat or very helpful. At the same time, 55 percent of Americans in a 2026 YouGov survey said AI is worse than human professionals at conducting therapy. Self-reported helpfulness is not the same as a measured clinical outcome.
Are these AI therapy statistics reliable?
Treat them as emerging rather than settled. Most figures come from individual surveys with modest sample sizes, different definitions of therapy versus general mental-health support, and self-selected respondents. The numbers are useful for spotting trends, such as rising use among younger people, but specific percentages can shift quickly as tools and awareness change.
Related AI therapy guides
References
- https://yougov.com/en-us/articles/54755-americans-are-increasingly-concerned-about-ai-exacerbating-mental-health-problems
- https://doi.org/10.1001/jamanetworkopen.2025.42281
- https://doi.org/10.1037/pri0000292
- https://www.apa.org/pubs/reports/chatbots-mental-health-2026
- https://yougov.com/en-us/articles/49480-can-an-ai-chatbot-be-your-therapist
- https://doi.org/10.1037/pst0000172