The growing popularity of AI chatbots is raising new concerns about youth mental health and suicide prevention. In this conversation, Christine Moutier, M.D., chief medical officer at the American Foundation for Suicide Prevention, discusses the usage of AI chatbots among adolescents, the potential dangers for young people experiencing mental health challenges, and the importance of educating clinicians and families about this new technology.
View Transcript
00:00:00:01 - 00:00:20:28
Tom Haederle
Welcome to Advancing Health. Increasing numbers of adolescents are turning to AI chatbots for mental support, a practice that experts say has the potential for both good and harm. What is AI's connection to suicide prevention, mental health and safety? We'll explore that question in this discussion.
00:00:21:00 - 00:00:42:29
Jordan Steiger
Welcome to Advancing Health. My name is Jordan Steiger, and I'm the director of Behavioral Health and Violence Prevention at the AHA. Really excited today to be joined by Dr. Christine Moutier, who is the chief medical officer at the American Foundation for Suicide Prevention. And we're going to have a great conversation today about youth mental health, suicide prevention and safety.
00:00:43:01 - 00:00:46:00
Jordan Steiger
So, Dr. Moutier, thank you for joining us today.
00:00:46:03 - 00:00:54:10
Christine Moutier, M.D.
Thank you so much for having me on. I'm very thrilled to reach the folks that your podcast reaches. So glad to be here.
00:00:54:13 - 00:01:14:18
Jordan Steiger
Yeah. So, you know, we're here to talk about, you know, something, I'm sure our listeners who care for children and adolescents and also the parents that are listening to this podcast, you know, something that they're probably thinking about all the time, which is the role AI is going to continue to play in their lives as we integrate AI more and more into our daily practices.
00:01:14:18 - 00:01:21:05
Jordan Steiger
And then, you know, like I mentioned before, kind of what are those connections to mental health and suicide prevention and safety?
00:01:21:07 - 00:01:39:19
Christine Moutier, M.D.
We're going to have so much to talk about because there's a lot going on within the clinical setting. But what we at AFSP have been even more focused on is actually out there in the general world of like general chatbot LLM utilization and seeing some of the things that are going on there.
00:01:39:25 - 00:01:57:13
Jordan Steiger
Yeah, I mean, I know that you and your team are doing so much work right now in this space, and I really think it's going to be valuable to our audience today. You know, the people that keep our hospitals and health systems running, both on the administrative side and the clinical side. You know, just really understanding this issue a little bit more.
00:01:57:16 - 00:02:13:06
Jordan Steiger
So could you just tell us what you are seeing out in the field when it comes to children and adolescents turning to AI chatbots for emotional support. And tell us maybe why these tools seem particularly appealing to kids who are experiencing mental health crises.
00:02:13:09 - 00:02:55:22
Christine Moutier, M.D.
Well, I would think of AI and the availability of chat bots and AI kind of services as being so ubiquitous now that we would need to actively create an environment that our children and others are growing up in and living, working, playing, to kind of curb that. And, and so that means that the impact on mental health and suicide as well, which we can look at both of those areas as complex health outcomes with like environmental, social drivers of risk, if you will, and protection in the suicide prevention field.
00:02:55:22 - 00:03:22:18
Christine Moutier, M.D.
We think of these secular forces in the world. And that's where I would put AI, and there's a lot of potential for good and there's potential for harm as well. And I think, you know, with regard to your question about why is this becoming such a thing? When people have mental health experiences and needs going on, it's not framed in their brain in that way.
00:03:22:21 - 00:04:07:13
Christine Moutier, M.D.
They're experiencing what they're experiencing. We are in a society that still has a lot of misunderstanding, obviously, around mental health experiences. And so finding a psychologically safe way to disclose what you're going through is something that we're all doing all the time. And I think particularly for people who are at that developmental stage where they haven't necessarily had practice unconsciously, the narrative and the dialog can go from very mundane, pragmatic, informational tool utilization to much more personal and health and mental health related dialog, even though the user may not be intending it that way.
00:04:07:19 - 00:04:33:09
Christine Moutier, M.D.
And we do have data that shows that people who have less access to health care are turning to chatbots for personal and mental health and social support reasons more than people who do have access. And that's been looked at by, you know, insured versus uninsured, or even when you look at marginalized populations and their issues with lacking access to care.
00:04:33:11 - 00:04:52:25
Jordan Steiger
I appreciate you bringing up that angle of this and that data, because that's something that, at the AHA we're really focused on is thinking about access to care. How do we create more access to care in our communities? And I can totally see what you're saying with this, where if you don't have that access, it's very easy to turn to this tool that's online.
00:04:52:28 - 00:05:13:15
Jordan Steiger
And it's, I think, an interesting point that you bring up that people might not necessarily be turning to an AI chatbot for mental health first. It might be just like a gradual thing. You know, you mentioned that there is kind of a line between this being a potentially very helpful resource and, you know, a tool that could inadvertently increase risk.
00:05:13:17 - 00:05:17:07
Jordan Steiger
So how do we kind of toe that line and what is that line.
00:05:17:10 - 00:05:42:25
Christine Moutier, M.D.
Yeah. Great question. And I'm happy to kind of try to talk about again why it's so important and then get into a little bit of breaking it down. So just kind of with both of your questions in mind and thinking about the fact that ChatGPT does now have, as an example, 1.2 billion weekly users, according to Open AI as of late fall 2026.
00:05:43:01 - 00:06:18:22
Christine Moutier, M.D.
And there's a recent study that looked at college students and found that those with moderate to severe symptoms of anxiety and depression were utilizing AI tools for mental health support at a much higher rate, double the odds of utilizing it in that way compared to the students without those symptoms and experiences. So it does mean that some of the more vulnerable among every population may be leaning on AI tools even more heavily related to mental health needs.
00:06:18:24 - 00:06:50:27
Christine Moutier, M.D.
And the reason that there is concern about that...some are data driven, some are observational, some are anecdotes or cases, you know, more known publicly available cases, media reports, lawsuits, and some are applying what we know about mental health clinically and from the science into this space. So just to say that there's sort of like varying degrees of what I'm going to say here in terms of how we know this.
00:06:50:29 - 00:07:26:06
Christine Moutier, M.D.
So there is the issue that there are some individuals who are prone to heavier utilization, almost in an addictive manner, where synthetic emotional intimacy and dialog that becomes very, very lengthy. That is a known phenomenon. And there are a number of issues with those types of users. And I will say, some of what I'm going to say about the harms are much, much more tilted on that heavy use end, whereas some of the harms can be there for really anyone who is experiencing mental health vulnerabilities even without that heavy utilization.
00:07:26:13 - 00:07:55:11
Christine Moutier, M.D.
So the heavy utilization is obviously going to, in a, you know, time in our lives zero sum game, it is likely going to detract from real human relationships, and especially for a young person or someone experiencing distress, that they may not be turning to a trusted adult or a mental health professional or any health professional, and instead to that AI LLM - that is problematic in, in and of itself.
00:07:55:13 - 00:08:29:12
Christine Moutier, M.D.
And then there are some of the ways that the dialogs are going that I think many of the companies are trying to improve these patterns that we're observing. But some of the patterns are around sycophancy, where the AI will endorse and applaud the person's thinking and ideas so much that for someone with vulnerabilities, it could reinforce cognitive distortions, hopelessness, feelings of intensity, of overwhelm, or even delusions and psychosis.
00:08:29:12 - 00:09:04:28
Christine Moutier, M.D.
Those are all known phenomenon that have happened, at least in case reports. There are also cases of AI LLMs giving suicide instructions or urging the user towards suicidal behavior. And sometimes that's because the individual asks them for those specific lethal methods. Sometimes it's that the user was speaking in metaphorical terms, and the AI may or may not have understood they were talking about taking their life, but instead, you know, coming home.
00:09:04:28 - 00:09:26:19
Christine Moutier, M.D.
Come home to me. Keep this between us. This kind of dialog can be very dangerous for someone who's in that suicidal mindset, where there are already cognitive distortions and that sort of tunnel vision that we know can happen neurophysiologically in the moments preceding a person's attempt or death.
00:09:26:22 - 00:09:49:05
Jordan Steiger
Absolutely. And that vulnerability, I think, is a huge piece of all of this. Thinking about, you know, hospital leaders and what they need to understand about this, and I know, Christine, you're a psychiatrist. You have done plenty of suicide screenings and risk assessments I'm sure, throughout your career. I'm a clinical social worker, same thing. You know, this is things that we're trained to look at and really understand.
00:09:49:06 - 00:10:03:29
Jordan Steiger
But do you think that, you know, in hospital based, you know, suicide screening, risk assessment, things like that - we should be talking to patients and families about AI LLMs and, you know, the safety aspects of all of this.
00:10:04:01 - 00:10:34:04
Christine Moutier, M.D.
I do, I think, especially for pediatricians or any youth serving health professional, that is a reasonable thing to do as a matter of just general history taking in the same way that I had recommended that social media utilization be incorporated into history taking, so that you can understand the patient's full world of experiences, kind of getting inside their internal world.
00:10:34:11 - 00:11:01:03
Christine Moutier, M.D.
I think it's important to educate parents and family members about just becoming more educated about what this means for us. There are chatbots now implanted in infant and toddler toys. And you know, when I first heard about this several months ago, I was kind of like, is this for real? And I had to look at it myself.
00:11:01:03 - 00:11:05:27
Jordan Steiger
And I just read an article about this myself. And I was also shocked. Yes.
00:11:05:29 - 00:11:32:11
Christine Moutier, M.D.
And they can be so sweet and innocent appearing. Right? If you think about what's going on with social and brain development and infants and toddlers and frankly, children all the way through young adulthood, there's a lot going on in development. We're allowing an influence, potentially a pretty heavy influence if the time is heavy, especially. But even if it's not, it's become part of their world.
00:11:32:11 - 00:11:34:23
Christine Moutier, M.D.
So I do have some caution around that.
00:11:34:27 - 00:12:02:23
Jordan Steiger
I mean, absolutely. And I think, you know, you mentioned educating patients and their families. I think really this starts with educating clinicians too. So this is a good first step today. And I know you are all doing a lot of work to do that, but just making sure that clinicians and administrators at hospitals really understand the risk and the need to talk about this without stigma, without judgment, which we know can be a big part of those conversations, unfortunately, sometimes.
00:12:02:25 - 00:12:32:24
Christine Moutier, M.D.
Yeah, it's so true. There's so many facets of mental health that warrant those first steps of clinician education. I mean, I'm thinking of even things like safety issues, like lethal means in the home where if we as clinicians aren't very comfortable with the topic, we have our own barriers then to raising it in a really, you know, the normal way we would in a compassionate, patient centered, clear, direct form of communication.
00:12:32:27 - 00:13:09:15
Christine Moutier, M.D.
I love the way you put it. Like that's the first step is for clinicians to feel a really solid sense of the latest, you know, ways of understanding, especially something that is emerging and rapidly evolving as this AI space is. When I trained now, I am a little bit getting into the older set. But, you know, training several decades ago, none of what we know now in terms of screening, risk assessment, safety planning, lethal means, counseling, most of that did not exist.
00:13:09:16 - 00:13:33:27
Christine Moutier, M.D.
And in fact, we were role modeled if not taught about contracting for safety and things that have been now proven to be ineffective and we don't recommend anymore. So it has actually been if you look at several decade time frame, a lot can change, obviously in every specialty: cancer, heart disease, I mean clearly emerging treatments, new ways of understanding.
00:13:33:27 - 00:13:39:07
Christine Moutier, M.D.
And that same thing is true for suicide screening and care.
00:13:39:10 - 00:13:59:03
Jordan Steiger
As we wrap up. You know, Christine, if you had a crystal ball and you were thinking about what this world of, you know, LLMs and youth, mental health and everything, you know, it's going to look like in the next five years or so, like, what should hospitals be thinking about? And how can they kind of maybe stay ahead a little bit of what you think is coming?
00:13:59:05 - 00:14:38:19
Christine Moutier, M.D.
I would say in the wider world of AI, I dream of a time where the companies would be working alongside with experts, including clinicians, scientists, people with lived experience embedded as part of the design team so that protective factors and experiences could be worked into the design, not just harm reduction. Certainly we need that. We need those safety benchmarks, but also be thinking about enhancing protective factors for mental health and suicide prevention.
00:14:38:22 - 00:15:21:27
Christine Moutier, M.D.
It's becoming an ambient force in our lives. So I do hope for that. And there are some reasons to hope, because there are many of us who are working within and alongside. Within the clinical environment, I think there are some really exciting things happening. I would call it in the machine learning, AI, precision medicine space of suicide research and suicide prevention, where the predictive modeling around the ability in a high risk population of patients to be able to say with the same level of precision that we can for other clinical outcomes outside of even outside of mental health, you know, the area under the curve is like the predictability score.
00:15:21:29 - 00:15:52:23
Christine Moutier, M.D.
And we're reaching now the .8.87 mark for area under the curve with 90 day suicide attempt predictability, again amongst a high risk population. That's very similar to the ability that other fields have to predict, you know, post-op outcomes and things like that based on risk factors and and how things went in the surgery. I think some of those will begin to be rolling out into practice.
00:15:52:23 - 00:16:22:00
Christine Moutier, M.D.
Right now, almost none of that exists. I would point health system leaders to the reach vet model at the VA. They have been using AI systems to be routinely analyzing about 140 factors in the electronic health record for veterans in terms of suicide risk, and when the suicide prevention coordinator or clinician sees the indicator rising, they reach out and call that veteran in.
00:16:22:03 - 00:16:49:07
Christine Moutier, M.D.
And that is a really proactive form of suicide prevention. That kind of modality is mostly being used in research, but again, is rolling out in some larger health systems even beyond the VA. If your health system has made suicide prevention a priority, then these are the ways, I think, to be staying fully on the cutting edge, to be kind of looking for, well, how would that be implemented actually, you know, in my health system.
00:16:49:13 - 00:17:13:16
Jordan Steiger
Yeah, those are some great points. And I think early intervention - any way we can achieve that - I think is going to be a true life saver for patients and family members. And I mean, hearing you today, I think I'm leaving this feeling a lot more hopeful about what the future can hold and how we can really use this to intervene and help kids and families and adolescents and everybody that touches their lives.
00:17:13:18 - 00:17:22:04
Tom Haederle
Thanks for listening to Advancing Health. Please subscribe and rate us five stars on Apple Podcasts, Spotify, or wherever you get your podcasts.



