Purpose A growing number of people are using LLMs for mental health support, including general-purpose AI, AI companions, and mental health LLMs. Reasons include privacy, 24/7 availability, perceived quality, and fear of judgment. These technologies raise questions about disengagement, harm, risk mitigation, accessibility, and discrimination. Current challenges align with the core competencies and values of peer support, yet there has been limited integration of peer supporters or people with lived experience in LLM research or implementation. Recent Findings Digital peer support training, certification, and core competencies already exist, but predate widespread LLM use. Digital navigators, including peer supporters, show promise in increasing digital self-determination and improving outcomes. While lived experience inclusion has gained traction, definitions and effective implementation practices remain unclear. Summary The foundations of peer support and evidence suggest that digital peer support and navigators can improve outcomes and increase education among those using LLMs. Leaders should prioritize expanding access to digital peer navigators and increasing LLM-related training for peer supporters. As technology continues to evolve, lived experience must be foundational to how we understand, design, and make decisions about these tools Opinion A growing number of people are using LLMs for mental health support, including general-purpose AI, AI companions, and mental health LLMs. Some reasons include viewing the tools as flexible, convenient, and effective. Others may use these tools to fill gaps in access to support and to avoid the fear of asking for help. Still others view them as alternatives, given mistrust of systems and a lack of culturally responsive resources and providers. While the growth of digital tools requires action to enforce accountability for quality and safety, many reasons people provide for using LLMs reflect foundational issues that led to the peer support movement. Issues such as voluntariness, person-centered support, harm reduction, and education are core to how peers support individuals in moving toward resources that meet their priorities and preferences. This paper does not argue that LLM use is inherently harmful, nor that peer support should replace digital tools. Instead, it argues that peer support is uniquely positioned to address the long-standing gaps that drive people to LLMs. Digital peer support and peer navigators can help people use, reduce, or disengage from LLMs in ways aligned with their own goals while strengthening connection to community and human support. More broadly, this moment calls for increased engagement of people with lived experience as experts to guide the shaping of these technologies. This requires conceptualizing lived experience not as a feedback mechanism but as epistemic justice, which centers those with direct knowledge to define what harms matter, what outcomes count, and what constitutes meaningful support.

Peer Support Principles and Navigation for LLMs in Mental Health

期刊论文
发表日期：

2026年3月31日

查看全文

Kelly Davis

摘要

Purpose

A growing number of people are using LLMs for mental health support, including general-purpose AI, AI companions, and mental health LLMs. Reasons include privacy, 24/7 availability, perceived quality, and fear of judgment. These technologies raise questions about disengagement, harm, risk mitigation, accessibility, and discrimination. Current challenges align with the core competencies and values of peer support, yet there has been limited integration of peer supporters or people with lived experience in LLM research or implementation.

Recent Findings

Digital peer support training, certification, and core competencies already exist, but predate widespread LLM use. Digital navigators, including peer supporters, show promise in increasing digital self-determination and improving outcomes. While lived experience inclusion has gained traction, definitions and effective implementation practices remain unclear.

Summary

The foundations of peer support and evidence suggest that digital peer support and navigators can improve outcomes and increase education among those using LLMs. Leaders should prioritize expanding access to digital peer navigators and increasing LLM-related training for peer supporters. As technology continues to evolve, lived experience must be foundational to how we understand, design, and make decisions about these tools

Opinion

A growing number of people are using LLMs for mental health support, including general-purpose AI, AI companions, and mental health LLMs. Some reasons include viewing the tools as flexible, convenient, and effective. Others may use these tools to fill gaps in access to support and to avoid the fear of asking for help. Still others view them as alternatives, given mistrust of systems and a lack of culturally responsive resources and providers.

While the growth of digital tools requires action to enforce accountability for quality and safety, many reasons people provide for using LLMs reflect foundational issues that led to the peer support movement. Issues such as voluntariness, person-centered support, harm reduction, and education are core to how peers support individuals in moving toward resources that meet their priorities and preferences.

This paper does not argue that LLM use is inherently harmful, nor that peer support should replace digital tools. Instead, it argues that peer support is uniquely positioned to address the long-standing gaps that drive people to LLMs. Digital peer support and peer navigators can help people use, reduce, or disengage from LLMs in ways aligned with their own goals while strengthening connection to community and human support.

More broadly, this moment calls for increased engagement of people with lived experience as experts to guide the shaping of these technologies. This requires conceptualizing lived experience not as a feedback mechanism but as epistemic justice, which centers those with direct knowledge to define what harms matter, what outcomes count, and what constitutes meaningful support.