How Many Youth and Young Adults Use AI for Mental Health Advice?

Summary: 5.4 million teens and young adults used AI for mental health advice in early 2025, based on data published in a new survey sponsored by Harvard University. While we only provide treatment for adults 18+ at Crownview Psychiatric Institute, this information can help us – and you – understand what to expect in the near future, and many participants in the survey will be adults sooner rather than later.

Key Points:

• Statistical analysis from the new survey shows 13.1% of teens and young adults use AI for mental health advice.
• The highest rates of use appeared among young adults 18-21.
• 65% of teens and young adults reported using AI for mental health advice at least once a month.
• 93% of users reported the advice they received was helpful.

Who Do Teens and Young Adults Ask for Mental Health Advice?

A growing body of evidence suggests that in the U.S., an increasing percentage of teens and young adults use AI for mental health advice. This is concerning because rates of diagnosed mental illness among teens and young adults steadily increased between 2010 and 2020, leveling off between 2020 and 2024. The increase in mental health disorders prompted to Surgeon General to declare a youth mental health crisis in the U.S., which affects all teens and young adults in the country, regardless of race/ethnicity, gender, economic status, or level of education.

To learn who teens and young adults ask for mental health advice, we’ll review data from a new publication called “Use of Generative AI for Mental Health Advice Among US Adolescents and Young Adults.”

Here’s how the researchers who completed this analysis describe the goal of the study:

“[To} report results from the first nationally representative survey of US adolescents and young adults aged 12 to 21 years examining the prevalence, frequency, and perceived helpfulness of advice from generative AI when feeling sad, angry, or nervous.”

Researchers conducted the study under the guidance of by the Harvard University Institutional Review Board (IRB), using the well-established STROBE protocol for guidance on reporting data from observational studies. Analysis allows statisticians to make population-level generalizations from the data on English-speaking youths in the U.S., between 12 and 21 years of age, with sufficient internet access to complete the survey.

We’ll share the results of this survey in a moment. First, we’ll share the general data that caused the Surgeon General to declare a Youth Mental Health Crisis, which will frame the size and scope of the topic we address in this article: how many teens and young adults use AI for mental health advice.

The most recent data on teen and young adult mental health appears in the 2025 National Survey on Drug Use and Health (2025 NSDUH), an annual survey conducted by the U.S. National Institutes of Health (NIH) in partnership with the Substance Abuse and Mental Health Services Administration (SAMHSA).

In the section below, we’ll share the facts and figures from the 2025 NSDUH on depression and anxiety among teens ages 12-17 and young adults ages 18-25.

Mental Health Among Adolescents and Young Adults

First, the latest data on depression among adolescents, as measured by the presence of a major depressive episode (MDE) and major depressive episode (MDE) with severe impairment, defined as a period of two weeks during which an individual experiences the primary symptoms of depression as identified by the Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition (DSM-5). The level of impairment refers to the degree to which symptoms disrupt typical daily function.

Here’s the data.

Past Two-Week MDE and MDE with Severe Impairment, Adolescents 12-17, 2024

• MDE, total: 15.1%
  • Female: 21.7%
  • Male: 8.7%
• MDE with severe impairment, total: 11.0%
  • Female: 16.0%
  • Male: 6.3%

Among adolescents with MDE, 58% of adolescents with MDE received some type of professional support from a human, while 62% of adolescents with MDE with severe impairment received some type of professional support from a human. This means close to 40% of adolescents who needed support for depression did not get the support they needed.

Next, the latest data on anxiety among adolescents, as measured by the presence of symptoms of generalized anxiety disorder (GAD) – mild, moderate, or severe – as defined as a period of two weeks during which an individual experiences the primary symptoms of GAD, identified by criteria established in the DSM-5, which we link to above. The level of severity refers to the degree to which symptoms disrupt typical daily function.

Past Two-Week Symptoms of Generalized Anxiety Disorder (GAD), Adolescents 12-17, 2025

• Mild symptoms, total: 23.6%
  • Female: 27.5%
  • Male: 19.9%
• Moderate symptoms, total: 10.6%
  • Female: 13.8%
  • Male: 7.6%
• Severe symptoms, total: 7.4%
  • Female: 11.0%
  • Male: 3.9%

No data available on rates of professional treatment and support among adolescents with GAD.

Next, we’ll share the parallel sets of data for young adults, also from the 2025 National Survey on Drug Use and Health (2024 NSDUH

Past Year MDE and MDE with Severe Impairment, Young Adults 18-25, 2025

MDE means major Depressive Episode

• MDE, total, 18-25: 14.2%
  • Male: 10.6%
  • Female: 17.8%
• MDE with severe impairment, total, 18-25: 10.2%
  • Male: 7.4%
  • Female: 13.1%

Among adults ages 18-25 with MDE, 59% received some type of professional support from a human, while 63% of adults 18-25 with MDE with severe impairment received some type of professional support from a human. As we observe among adolescents, this data shows that close to 40% of adults ages 18-25 who needed support for depression did not get the support they needed.

Past Two-Week Symptoms of Generalized Anxiety Disorder (GAD), Young Adults 18-25, 2025

• Mild symptoms, total: 21.1%
• Moderate symptoms, total: 7.8%
• Severe symptoms, total: 4.3%

No data available on rates of professional treatment and support among adolescents with GAD.

That’s the current situation with regards to youth and young adult mental health in the U.S. Although the upward trend in prevalence rates has stabilized over the past several years, the fact remains that millions of youth and young adults have serious mental health issues, with less than 40 percent receiving the professional mental health support they need.

Now let’s look at the nationwide survey we introduce in the beginning of this article – “Use of Generative AI for Mental Health Advice Among US Adolescents and Young Adults” – and learn where teens and young adults go for support, and find out how many teens and young adults use AI for mental health advice.

Who Used AI for Mental Health Advice? About the Study

Here’s an overview of the people who took the survey. We include this information to verify its diversity, as compared to other surveys of its type, which skew heavily towards participants from white families whose parents have a college education.

Survey participants included:

• 1058 teens and young adults, age 12-21
  • Female: 53%
  • Male: 47%
• By age:
  • 12-14: 41%
  • 15-17: 45%
  • 18-21: 14%
• By race/ethnicity:
  • Black: 10%
  • Hispanic: 24%
  • White: 54%
  • Other: 12%
• Parental education:
  • High School diploma: 29%
  • Some college: 30%
  • Bachelor’s degree: 41%

That’s the demographic breakdown. It’s not a perfect representation of the U.S. population, but it’s a fair cohort from which we can draw valid conclusions.

The survey asked participants the following questions, with the following possible answers:

1. Have you ever used generative AI?
   • Yes / No
2. Have you sought advice or help – i.e. mental health advice – from generative AI when feeling sad, angry, or nervous?
   • Yes / No
3. How often do you seek mental health advice from generative AI?
   • Never / Hardly ever / Once a month or more / Once a week or more / Daily or almost daily
4. Was the mental health advice you received from generative AI helpful?
   • Not helpful /Somewhat helpful / Very helpful

Let’s take a look at what they found.

How many Teens and Young Adults Use AI for Mental Health Advice? The Results

We’ll share the survey outcome by age group, starting with teens age 12-14.

Age Group 1, Ages 12-14

How many 12–14-year-olds used AI for mental health advice?

• Total: 13%
  • Female: 8%
  • Male: 5%

How often did they use AI for mental health advice?

• Frequency:
  • Never: 4%
  • Hardly ever: 22%
  • Once a month or more: 39%
  • Once a week or more: 33%
  • Daily or almost daily: 3%

What did they think about the advice they received?

• Helpfulness:
  • Not helpful: 8%
  • Somewhat helpful: 61%
  • Very helpful: 29%

Now we’ll look at the same data for 15-17-year-olds.

Age Group 2: 15-17

How many 15-17-year-olds used AI for mental health advice?

• Total: 21%
  • Female: 16%
  • Male: 5%

How often did they use AI for mental health advice?

• Frequency:
  • Never: 4%
  • Hardly ever: 20%
  • Once a month or more: 38%
  • Once a week or more: 35%
  • Daily or almost daily: 6%

What did they think about the advice they received?

• Helpfulness:
  • Not helpful: 6%
  • Somewhat helpful: 74%
  • Very helpful: 20%

Next, the same data for 18-21-year-olds.

Age Group 3: 18-21

How many 18-21-year-olds used AI for mental health advice?

• Total: 48%
  • Female: 19%
  • Male: 29%

How often did they use AI for mental health advice?

• Frequency:
  • Never: <1%
  • Hardly ever: 39%
  • Once a month or more: 33%
  • Once a week or more: 11%
  • Daily or almost daily: 17%

What did they think about the advice they received?

• Helpfulness:
  • Not helpful: 7%
  • Somewhat helpful: 52%
  • Very helpful: 41%

We’ll discuss these results below.

Millions of Young Adults Use AI for Mental Health Advice. But Should They?

That’s an easy question to answer:

No, young adults should not use AI for mental health advice. In fact, no one of any age should use AI for mental health advice.

Why?

In a nutshell, because data shows they do not offer accurate, evidence-based provide answers consistently enough to be considered safe and reliable.

How do we know?

We have data that shows AI chatbots for mental health are neither safe nor accurate.

First, a group of researchers conducted and published a study that shows the following:

• AI provided safe and accurate answers to mental health questions only 80% of the time.
• Human therapists answered mental health questions safely 93% of the time.

In a recent study – “Characteristics and Safety of Consumer Chatbots for Emergent Adolescent Health Concerns” – a group of researchers assessed answers offered by two types of AI chatbots about menta health. They determined the percentage of answers that met minimum criteria for the following: appropriateness, empathy, understandability, appropriate referral, appropriate escalation of support.

What they found is both frightening and informative.

AI Answers to Mental Health Questions

Companion Bots:

• Appropriate answers: 22%
• Empathetic answers: 42%
• Understandable answers: 71%
• Referred individual to appropriate resource: 11%
• Appropriately identified need for escalation: 40%

General Assistant (GA) Bots:

• Appropriate answers: 83%
• Empathetic answers: 93%
• Understandability: 97%
• Understandable answers: 73%
• Appropriately identified need for escalation: 90%

These answers are informative in that they show us that general assistant bots offer answers that are safe and accurate with far greater frequency than companions bots. There’s a clear reason: companion bots are sycophantic by design. They provide answers that do two things: increase the likelihood an individual will ask a follow-up, make the individual feel good about themselves. In contrast, general assistant bots are designed to provide accurate information, rather than companionship. This increases the default level of accuracy of GA bots.

Now let’s discuss why the results of this study – and the one we focus on throughout this article – are frightening as well as informative.

How We Can Use This Data: Increase Engagement With Young People on Mental Health Topics

Our primary takeaway here is important: we need to normalize conversations about mental health and mental health topics among teens and young adults. When we review the data on usage and compare it against the data on accuracy, appropriateness, and safety, we realize we have a problem.

Let’s take a basic analysis of how chatbots do, with regards to accuracy, appropriateness, and safety when answering mental health questions:

They don’t do well at all. They regularly provide answers that fail to meet the most basic criteria for good therapy. Answers are not consistently safe, appropriate, or accurate.

Now let’s consider what teens and young adults thought about the answers they received to mental health questions:

Over 90 percent considered the answers ‘very helpful.’

That’s the frightening part. Most teens and young adults considered the answers they received as very helpful, without understanding that the answers they received may have been, in fact, very harmful. To change this, here’s what we can do:

1. Talk about mental health early and often.
2. Give youth real, evidence-based answers to their questions when they ask them.
3. Ensure teens and young adults understand the basics of mental health safety.
4. Create an atmosphere – at home, at school, and across society – where teens and young adults turn to humans for mental health advice first, rather than AI.
5. Model openness to and acceptance of evidence-based treatment for mental health.

When we do that, and lead with empathy and compassion, we can change our cultural milieu from one where young people are more comfortable asking AI bots for mental health advice to one where young people feel more comfortable asking a parent, a counselor, or a mental health professional for mental health advice.

That’s how we can keep our teens and young adults safe: create an atmosphere that increases the likelihood they’ll come to us for advice about their mental health challenges, rather than an AI chatbot.