Ottawa researchers take on the AI therapist problem
University of Ottawa researchers are joining a growing wave of academics trying to figure out whether an AI chatbot can safely function as a mental health support tool — and what happens when it can't. The push comes amid rising concern over what's being informally dubbed "chatbot psychosis," where people using largely unregulated AI companions for emotional support have reportedly spiraled into delusional or disordered thinking.
Why this matters for Ottawa
As a city with a major university research hub and a growing tech sector in Kanata North, Ottawa is well positioned to be part of the conversation shaping how AI mental health tools get built — and regulated — in Canada. uOttawa's involvement puts the capital directly in the middle of a debate that's playing out globally: can chatbots, built without clinical oversight, be trusted with vulnerable users' mental health? For Ottawa residents who might turn to free AI apps for a listening ear during a tough moment, the stakes are personal, not just academic.
The core concern: an unregulated AI landscape
According to reporting from the Ottawa Citizen, the current AI chatbot ecosystem is minimally regulated, and that gap is what researchers say is contributing to real harm. Without clinical guardrails, chatbots can reinforce a user's distorted thinking rather than gently challenge it — the opposite of what a trained therapist is taught to do. Some users engaging in long, emotionally intense conversations with general-purpose AI chatbots have reportedly developed delusional beliefs, sometimes reinforced by the AI's tendency to agree with or elaborate on whatever the user brings to the conversation.
What researchers are trying to build instead
Rather than simply warning people away from AI chatbots, uOttawa's research effort is aimed at the harder problem: figuring out what a safe AI therapist would actually look like. That means examining how a chatbot should be designed to recognize warning signs of psychological distress, avoid validating harmful delusions, and know when to redirect a user toward real human care rather than continuing the conversation. It's a technical and ethical challenge at once — the kind of interdisciplinary work universities are well suited to tackle, pulling together computer scientists, psychologists, and ethicists.
Part of a bigger, global reckoning
uOttawa isn't alone in this. Researchers at institutions around the world are racing to study the mental health risks of AI companionship apps as their popularity has exploded, particularly among people seeking cheap or immediate emotional support outside the formal healthcare system. That demand isn't going away — access to in-person mental health care remains difficult in many communities, including parts of Ottawa, which is part of why AI tools have filled the gap so quickly, for better or worse.
What to watch next
It's early days for this kind of research, and there's no indication yet of a finished, publicly available "safe" AI therapy tool coming out of uOttawa. But the university's involvement signals that Ottawa will be part of shaping how these tools evolve — and potentially how they get regulated — in the years ahead. For now, mental health professionals continue to caution that AI chatbots, however conversational they seem, are not a substitute for licensed care.
Source: Ottawa Citizen
