How NAMI Is Helping Push For Clarity and Safety in AI Mental Health Tools
Led by science, shaped by lived experience
AI is becoming part of how people look for information and support, including around mental health, but the quality of what they receive can range from helpful to confusing to unsafe. New polling highlights how urgently clarity and safeguards are needed. A recent NAMI/Ipsos survey shows that 12% of adults are likely to use AI chatbots for mental health care in the next six months, with 1% of adults saying they already do. As AI does not represent clinical expertise, it is important for people to understand how these tools behave so they can make informed decisions about their own care.
These findings reinforce the need for clear, independent information about how these tools behave when people look for help. And right now, there is no trusted way for the public to understand those differences. So that’s why NAMI, the nation’s largest grassroots mental health organization, is stepping in to bring clarity, safety, and the voice of lived experience to this moment.
We are taking a careful, long-term look at how AI tools respond when people ask questions about mental health. Our goal is simple: give people clear, trustworthy information as AI becomes more present in their lives, without endorsing any tool, replacing human support, or suggesting AI is ready for clinical use.
To do this work, NAMI is partnering with Dr. John Torous, director of Digital Psychiatry at Beth Israel Deaconess Medical Center, a Harvard Medical School–affiliated teaching hospital and a national leader in digital mental health research.
What We’re Doing
We will examine how AI tools behave when people turn to them for mental health information, including whether they:
- Recognize safety concerns and offer appropriate next steps
- Provide accurate, evidence-informed information
- Respond in respectful, supportive, and inclusive ways
- Avoid implying privacy protections or encouraging unsafe personal disclosures
- Stay within safe informational boundaries, rather than acting like therapy
To do this, the team is:
- Creating realistic, everyday scenarios based on how people actually use AI tools
- Collecting AI-generated responses
- Having clinicians and people with lived experience review them for safety, accuracy, supportiveness, respect, and privacy awareness
- Involving NAMI leaders, peers, families, volunteers, clinicians, and researchers
What to Expect
This work will offer:
- Clear, easy-to-understand information about how AI tools respond
- Insight into strengths, gaps, and potential risks
- Support for informed decision-making (without implying AI is ready for clinical use)
- Independent feedback developers can use to improve their tools
- A meaningful role for NAMI communities in defining expectations for safe, supportive AI
This work will offer:
- Clear, easy-to-understand information about how AI tools respond
- Insight into strengths, gaps, and potential risks
- Support for informed decision-making (without implying AI is ready for clinical use)
- Independent feedback developers can use to improve their tools
- A meaningful role for NAMI communities in defining expectations for safe, supportive AI
Timeline
The work is in its early stages and findings will be shared in phases as the work progresses and as technology evolves.
FAQ: NAMI’s Work on AI and Mental Health
NAMI is the largest grassroots mental health organization in the United States, and our nationwide lived experience community gives us both the responsibility and the credibility to help bring clarity and safety to AI in mental health. People are already using AI to ask questions about mental health, but the information they receive can range from helpful to confusing to unsafe, and there is no trusted way for the public to understand these differences. That is why NAMI is stepping in to develop independent benchmarks that help people understand what different tools can and cannot do. We will never endorse specific AI products, but we will provide clear, trustworthy information people can use to make their own choices.
Dr. Torous and the Division of Digital Psychiatry at Beth Israel Deaconess Medical Center are leaders in the field of digital mental health. Dr Torous and his team support mindapps.org (the largest database of mental health apps) and the Digital Navigator program to promote digital equity and access. Dr. Torous has published over 300 peer-reviewed articles and five book chapters on the topic of digital mental health. Dr. Torous is also part of NAMI’s Scientific Advisory Council, providing guidance and expertise on emerging areas of scientific and research advancement. He chose to work with NAMI because our nationwide lived experience community brings an essential perspective on what safe, trustworthy support must look like.
A recent NAMI/Ipsos poll* shows that more than one in ten adults (12%) are likely to turn to AI chatbots for mental health treatment or therapy in the next six months. As AI does not represent a substitute for clinical care, these findings underscore the need for clear, independent information about how these tools behave when people look for help, which is why NAMI is undertaking this work.
*The NAMI/Ipsos poll was conducted November 7–9, 2025 using Ipsos’ probability-based
KnowledgePanel® and is based on a nationally representative sample of 2,046 adults age 18 or older.
The survey has a margin of error of ±2.3 percentage points.
The early phases of the project will look at three core areas that reflect what people with lived experience tell us matters most when they look for support:
- Safety and crisis response.
Whether tools recognize when someone may be in distress and offer appropriate, safe next steps.
- Accuracy and quality of information.
Whether responses are factually correct, consistent with evidence, and free of harmful or misleading claims.
- Cultural relevance and human support.
Whether AI tools respond respectfully across cultures, identities, and lived experiences and whether they offer supportive, human-centered language.
No. NAMI and Dr. Torous are partnering to develop benchmarks that help people understand how different AI tools respond when asked about mental health.
No. NAMI does not endorse AI for mental health treatment, for any age group or any mental health condition. AI may help people access general information or resources when the tools are developed safely and responsibly, but it is not a replacement for care. This is why clear, independent benchmarks are needed.
AI will not replace human connection. This project is entirely distinct from the HelpLine, and all NAMI HelpLine operations will remain the same, with trained volunteers and staff available 10 AM to 10 PM EST, Monday through Friday.
No. NAMI programs, presentations, support groups, and outreach activities remain central to how we support people and families. AI will not replace these efforts.
The benchmarks may be shared by NAMI State Offices and local NAMI affiliates, and as NAMI content is updated over time, information about AI and mental health may be integrated where it enhances participant understanding and support.
NAMI National staff, NAMI State Office and NAMI affiliate staff and volunteers, family members and peers, and the Division of Digital Psychiatry at BIDMC will all play a role in creating the benchmarks for this project.
Lived experience is central to this work. Peers, family members, volunteers, and affiliate leaders will help shape the scenarios, identify what “safe and supportive” looks like, and review AI-generated responses. This ensures the benchmarks reflect the realities of how people actually seek information and support.
The benchmarks will allow people to look at a specific aspect of mental health support, such as whether an AI tool provides accurate information about crisis lines, and compare how different tools perform. Individuals can then determine what matters most to them and make informed decisions based on clear, accessible data.
Responsible AI developers can use the information from this unbiased and expert evaluation to improve their products and make sure they provide safe, accurate, and appropriate information.
