New Tool Evaluates Mental Health Advice in AI Chatbots

Researchers developed a new tool to assess the quality of mental health guidance provided by major AI chatbots.

Updated on Oct. 30, 2026 in Teens

Bold vector editorial illustration of translucent geometric cubes and prisms, representing an analytical tool for evaluating mental health data.
The RAND Corporation has introduced a new diagnostic tool designed to measure the quality of mental health guidance offered by AI-powered chatbots. AI Illustration. Upload story photo >

Live Poll

Do you trust artificial intelligence chatbots to provide accurate mental health advice to teenagers?

The RAND Corporation has released the Adolescent Mental Health AI Response Tool to measure the quality of advice chatbots offer to teens. This new system evaluates how platforms perform across seven distinct clinical domains.

Why it matters

Adolescents are increasingly turning to general-purpose AI chatbots for mental health guidance. However, these tools often lack consistent evaluation against clinical standards, creating a need for objective performance metrics.

The new tool assesses AI performance across seven domains, testing four major models including ChatGPT-5, ChatGPT-4, Claude 4.5, and Gemini 2.5. Initial scoring shows that advice for anxiety is more comprehensive than for depression.

The players

RAND Corporation

This is a nonprofit global policy think tank that conducts research and analysis on complex public policy challenges.

The details

Clinicians helped develop the tool by reviewing items for clinical importance and objective validity. An ensemble of three large language models then scored multiturn conversations to determine how effectively the AI handled mental health queries.

Timeline

  1. October 30, 2026: The research results were published.

Culture Shift

This research follows the established precedent of applying institutional rigour to evaluate emerging mental health technologies. It signals a move toward systematic oversight as adolescents increasingly rely on digital tools for wellness support.

Teens and parents should be aware that AI chatbots may offer inconsistent advice depending on the severity of the mental health concern. Users should treat AI responses as supplemental information rather than a replacement for professional clinical care.

The takeaway

AI models currently struggle to provide high-quality guidance for severe depression compared to anxiety and moderate conditions. Families should prioritize consulting licensed professionals for any serious mental health needs rather than relying solely on chatbot responses.

Further reading

Learn more about the latest digital wellness trends for the younger generation in our Teens section.

More information

View the complete findings on the RAND research publication page.

Source note: This article includes information reported by Rand Corporation.

Live Poll

Do you trust artificial intelligence chatbots to provide accurate mental health advice to teenagers?