AI is already reaching people in distress, but mental health safeguards lag behind
· Medical Xpressedited by Swati Mestri, reviewed by Robert Egan
Swati Mestri
Scientific Editor
Meet our editorial team
Behind our editorial process
Robert Egan
Senior Editor
Meet our editorial team
Behind our editorial process Editors' notes
This article has been reviewed according to Science X's editorial process and policies. Editors have highlighted the following attributes while ensuring the content's credibility:
fact-checked
peer-reviewed publication
trusted source
proofread
The GIST Add as preferred source
As AI use gains traction in mental health care, Deakin University researchers have raised the alarm that there are currently no agreed-upon frameworks in place to regulate or guide its application.
The team from Deakin's Lifespan Institute has published a position paper in The Lancet Psychiatry examining the ways people experiencing mental health concerns are coming into contact with AI. The paper outlines areas where new regulations, guidelines and other protocols should be introduced to monitor the use of AI and ensure it is implemented safely and effectively.
Associate Professor Jake Linardon said the purpose of the study wasn't to demonize AI, but rather to ensure it is implemented correctly in such a high-stakes space.
"The possibilities AI brings to the mental health space are significant, including the potential for faster and more accurate diagnosis, more tailored treatment and ultimately better patient outcomes."
"However, as we are all finding out these past few years, AI has been introduced with such speed that we are using it without really understanding it. This has significant implications for people experiencing mental illness, and it also means we shouldn't automatically view AI as the answer to all our problems," Linardon said.
The benefits of AI during the health care journey
Linardon said that rather than replacing clinicians, there are lots of ways AI can enhance care—including making mental health care more tailored and, in some cases, more accessible.
"Imagine a young person presents to a psychologist with worsening anxiety, poor sleep, low mood and trouble functioning at work or study. AI might help at several points along that pathway.
"Early on, it could support assessment by helping clinicians pull together patterns from questionnaires, symptom histories or even passive data from a smartphone app, such as changes in sleep, activity or social withdrawal," Linardon said.
"Later, it might assist with documentation, summarizing a session so the clinician can spend less time writing notes and more time focusing on the person in front of them."
"AI could also support treatment planning by highlighting patterns that suggest who may need more intensive follow-up or who may be responding well to a lower-intensity option," Linardon said.
On the treatment side, AI can extend support between appointments in the form of a smartphone tool that checks in on symptoms, helps the person practice coping skills at difficult moments or flags early warning signs that they may be deteriorating.
In the ideal scenario, it may mean care becomes more responsive and more personalized, rather than relying only on brief snapshots taken during appointments.
"The key point of our paper is, while there are certainly benefits to be had, it is important not to simply assume each of these touchpoints is working as it should," Linardon said.
When might AI become a problem in mental health care?
The paper identifies that the same technology that helps with monitoring or support could also create problems if it's inaccurate, poorly governed or used without enough oversight.
AI in this setting isn't a single thing but can include patient-facing tools, clinician-facing tools and system-facing applications, each raising different ethical, clinical and regulatory issues.
"For example, if an AI system gives unsafe advice, misses a suicidal disclosure, overstates risk or subtly changes what a patient chooses to share—that has real clinical consequences," Linardon said.
"It could also shift the therapeutic relationship if patients start seeing the AI as more available, more responsive or even more authoritative than the clinician.
"That is why we argue that the process needs strong safeguards from the outset.
"In practice, that means proper evidence that tools actually help, much better safety testing, clear escalation pathways for high-risk situations, transparency about what the AI is doing and clear rules about who remains responsible for decisions.
"It also means involving patients in the design and evaluation of these tools, so the technology is built around real needs rather than assumptions from developers alone.
"The field is at a point where the foundations really matter. If the evidence, safety standards and governance are weak, then introducing AI into clinical mental health care too quickly could do more harm than good."
How best to work with AI moving forward
Linardon and his team at the Deakin Lifespan Institute have put forward a practical roadmap to help guide the optimal use of AI in mental health care.
It includes a particular focus on clarifying the optimal sequencing and integration of AI tools into clinical workflows for health care professionals.
The paper highlights the importance of including general-purpose AI systems, such as OpenAI's ChatGPT, Anthropic's Claude and Google's Gemini, together with evaluation of dedicated purpose-built mental health apps that incorporate AI.
Including general-purpose, public-facing AI systems as part of the overall picture of patient care acknowledges and accounts for the fact that they are often the patient's first point of contact.
"The most urgent priorities over the next few years are ensuring safety, evidence, transparency and equity. Other steps, like widespread clinician training or large-scale health system integration, are important, but they depend on first having tools that are clinically useful and properly governed.
"For people experiencing mental illness, these findings matter because AI tools are already being used in moments of distress."
"In practice, the findings point to a more cautious and structured rollout of AI in mental health care. Rather than rushing tools into clinics or services, they suggest we first need better evidence and stronger safety checks," Linardon said.
"Once that foundation is in place, AI could be applied more responsibly to areas like clinical documentation, decision support and service delivery, without asking clinicians or patients to rely on tools that have not been properly tested."
Publication details
Jake Linardon et al, Multidisciplinary research priorities for artificial intelligence in mental health: a call to action, The Lancet Psychiatry (2026). DOI: 10.1016/s2215-0366(26)00127-6
Journal information: The Lancet Psychiatry
Key medical concepts
Clinical categories
PsychiatryPsychology & Mental health Provided by Deakin University Who's behind this story?
Swati Mestri
Swati Mestri holds a bachelor's degree in Electronics Engineering and has worked as a content editor since 2019. She has experience editing research documents across technology, health care, and materials science, and has a particular interest in technology and space. Full profile →
Robert Egan
Bachelor's in mathematical biology, Master's in creative writing. Well-traveled with unique perspectives on science and language. Full profile →
Citation: AI is already reaching people in distress, but mental health safeguards lag behind (2026, August 14) retrieved 14 August 2026 from https://medicalxpress.com/news/2026-08-ai-people-distress-mental-health.html This document is subject to copyright. Apart from any fair dealing for the purpose of private study or research, no part may be reproduced without the written permission. The content is provided for information purposes only.