Psychiatrists propose new framework to screen patients for AI use

A new framework aims to help mental-health professionals screen patients for their use of artificial intelligence, such as chatbots, during clinical assessments.

The AWARE tool was developed by a pan-Canadian research team led by Dr. Alexandre Hudon, a psychiatrist and clinical professor at Université de Montréal.

"We wanted to figure out how to integrate discussion of AI use into our clinical practice, just as we ask about smoking, drugs and so on," said Hudon.

READ MORE: Peel paramedics launch medical drone pilot project to deliver life-saving supplies in Bolton

READ MORE: Canadian women travel to Mexico for endometriosis surgery amid long waits at home

The acronym stands for the five dimensions it examines: AI use, Why, Attachment, Reality and risk, and Effect on functioning.

The goal is to understand how often patients use AI, what tools they use and their motivations.

"One thing that makes talking to an AI model different is that it leads to an interaction," Hudon noted.

"The machine responds, sometimes remembers the context of a conversation and can adopt a warm or empathetic tone. As a result, an emotional investment in the relationship can develop."

The framework also looks at how AI influences a person's perception of reality and decision-making.

"Imagine a person who believes they are being watched by the people around them," Hudon said.

"They ask a chatbot to help them interpret what they are experiencing. Depending on how the system responds, the conversation could reinforce their perception or lead them to question it."

Finally, it considers the impact on daily functioning, such as sleep, work, studies or social relationships.

Hudon described AWARE as a practical framework that can complement a psychiatric interview, not a checklist for diagnosing an AI addiction.

The approach is deliberately non-confrontational.

"A person may think of their conversations with an AI model as something very private, or feel uncomfortable talking about it," Hudon explained.

"A series of questions that sound accusatory could backfire and cause the person to hide their use of AI. The purpose of our approach is screening."

The research team believes discussing AI should become part of best practices for mental-health professionals.

"Rather, the point is to recognize that AI has become a part of many patients’ daily lives, and it can now play a role in how they think, seek information, manage their emotions or maintain their relationships," Hudon said.

Among the potential risks, he cited an increase in depressive or anxiety symptoms linked to AI-amplified concerns, as well as changes in sleep patterns.

The framework is a first attempt to integrate the impact of AI into clinical practice across Canada.

Hudon stressed that several steps remain, including validating the framework and evaluating its implementation in healthcare settings.

The research was published September 16, 2026 in JMIR Medical Education.

With files from Université de Montréal