| Journal of Community Safety and Well-Being (2026) 11(3), 163–165. | https://doi.org/10.35502/jcswb.586 |
Over the past decade, artificial intelligence (AI) has been repeatedly recognized as a revolutionary tool for enhancing human efficiency (Balan & Gumpel, 2025). A significant recent development in the AI field is the rise of conversational systems that are widely accessible to the public. These systems are based on large language models (LLMs), with their most well-known product being ChatGPT, developed by OpenAI (OpenAI, Inc./OpenAI Global LLC, San Francisco, CA, USA) (Balan & Gumpel, 2025). ChatGPT is built on the generative pre-trained transformer (GPT) model, which enables it to understand and interpret user requests. This understanding allows it to generate appropriate responses in natural human language (Alawida et al., 2023). The GPT employs a wide range of generative and discriminative techniques to produce these responses by learning from vast amounts of data (Alawida et al., 2023). To put it simply, we have never seen the size and advancement in training in previous language models before (Alawida et al., 2023). This innovation also sparks curiosity within various communities about what else remains to be learned about the accessibility of ChatGPT, and, more importantly, which parts of society are most susceptible to this enhancement in cognitive processing with the support from AI.
The authors of this commentary are mental health professionals who are both curious and concerned about the rapid integration of AI in our everyday world – especially its use among youth for mental health. The purpose of this commentary is to provide an overview of the current AI landscape, concerns from evidence-based literature, the impact on youth cognitive processing and empathy, and the increased use of AI for youth mental health. We end this commentary with an inquisitive lens and a call to our colleagues – how can we continue to learn about AI and its impact on general safety and safety for vulnerable populations?
Research supports that ChatGPT has over 100 million users (Nazir, 2026). We aim to explore and discuss the use of ChatGPT and chatbots among youth, with a focus on its use for mental health support. We will also consider the ethical and practical implications of its role in digital emotional intelligence and human connection simulation. Recent Canadian research suggests that young people view AI as a useful supplement for mental health information and support, although concerns remain regarding privacy, safety, and the importance of maintaining human-centred care (Ding & Barbic, 2025; Mitchell & Ameis, 2025).
Considering adolescents as the cornerstone of social, biological, and psychological development, their neurological development is now being paralleled by AI. AI is seen as a revolutionary tool to enhance human function, but can something be enhanced before it is fully developed? Growing up in environments filled with constant digital immersion, early exposure to algorithm-driven tools, and high engagement with interactive technologies, adolescents and young adults have developed a unique digital fluency (Giansanti & Cosenza, 2026).
Generative AI has been integrated into the daily practices of young people in Canada more rapidly than in any other age cohort. In academic contexts, 73% of Canadian post-secondary students now report using generative AI for coursework, an increase from 59% in 2024 and 52% in 2023. Notably, 48% of users perceive a decline in their critical thinking abilities since adopting these tools (KPMG Canada, 2025). The use of generative AI extends beyond academic work. In roundtables organized by McGill University’s Centre for Media, Technology and Democracy and Simon Fraser University’s Dialogue on Technology Project, participants aged 17–23 years described chatbots as habit-forming, reported emotional reliance, and noted a loss of independent brainstorming, all of which were described as difficult to reverse. Participants attributed these effects to intentional design, asserting that the tools are engineered to encourage repeated use (The Canadian Press, 2026). As a result, youth adoption of generative AI is both widespread and subject to critical reflection by young people themselves.
Curiosity arises when we consider the development of digital fluency within youth alongside the research that indicates that cognitive and affective empathy develop in childhood and mature during adolescence, continuing into adulthood. This notion is supported by the theory of mind which allows for a cognitive understanding of others’ thoughts and intentions (Jacob & Reddy, 2024). Improved social cognitive skills can enhance the neural activity of the empathy network. However, does the digital fluency of chatbots facilitate that improvement in almost every social cognitive skill? Changes in social brain development during adolescence are responsible for the appropriate detection and growth of empathy. These changes impact various neural networks and social cognitive processes (Jacob & Reddy, 2024). Major reviews explain adolescent neurocognitive development through the relationships between regulatory and affective cognitive processes and their associated neural circuit.
There are inherent risks in these technologies, which may result in inadvertent negative consequences for youth seeking mental health support. Based on an 18-month ethnographic collaboration with licensed psychologists and trained peer counsellors, Iftikhar et al. (2025) mapped the conversational behaviour of LLMs against professional codes of conduct and identified 15 distinct categories of ethical risk. These include mishandling of crisis situations, reinforcement of users’ negative beliefs, unfair discrimination, lack of contextual adaptation, and what the authors term deceptive empathy: the generation of language that simulates understanding without genuine comprehension, accountability, or clinical judgment. Moore et al. (2025) reported similar findings, noting that widely used models express stigma toward individuals with mental health conditions and respond inappropriately to clinically critical situations, partly because AI systems tend to validate rather than challenge user assertions. Since these models are trained on text that encodes prevailing social stratifications, their outputs reproduce these biases. Consequently, the Canadian Mental Health Association (CMHA, 2026) has cautioned that chatbot responses may reinforce racist and misogynistic content for equity-seeking users. In addition to bias, these systems are not designed for the cultural contexts of the youth who use them. Systems trained primarily on English-language and American-centred data are poorly equipped to address the needs of Indigenous, Francophone, newcomer, and rural youth in Canada. The lack of culturally appropriate treatment is already among the most frequently identified barriers in Canadian youth mental health literature (Kourgiantakis et al., 2023). Technologies that cannot adapt to context perpetuate barriers to comprehensive care while appearing responsive. Developmental factors further compound these risks, as younger users are less likely than adults to challenge model outputs, resolve misunderstandings, or identify errors, and their still-developing self-regulation increases susceptibility to emotionally reinforcing interactions.
The adoption of AI for mental health support among Canadian youth is influenced by both the lack of accessible mental health services and the ease of access to technology. The Canadian Institute for Health Information (2025) reports that 36% of children and youth aged 2–17 years with a perceived need for mental health care had needs that were partially or completely unmet, increasing to 40% among girls, and peaking at 52% among young adults aged 18–34 years. Research from Ontario indicates that youth aged 16–24 years experience the highest rates of mental health and addiction concerns, as well as the greatest unmet need. Principal structural barriers include limited service availability, cost, prolonged wait times, fragmentation, discontinuity across the child-to-adult transition, stigma, racism, and a lack of culturally appropriate treatment (Kourgiantakis et al., 2023). Technologies that are free at the point of use, continuously available, theoretically non-judgmental, and accessible without referral are therefore filling a critical gap in the system. However, these AI systems are not subject to the same regulatory oversight as other mental health services.
Furthermore, the substitution of AI for traditional mental health support is occurring without the safeguards present in other domains of practice. As noted by the SickKids Child Health Policy Accelerator (2026) in a brief to the Standing Senate Committee on Social Affairs, Science and Technology, mental health services in Canada are provided by regulated professionals, yet there is no comparable oversight for mental health guidance delivered by AI chatbots and companions. This regulatory gap is particularly concerning given evidence that Canadian adolescents are presenting serious concerns, including non-suicidal self-injury impulses and suicidality, to systems that demonstrate inconsistent performance in crisis situations.
Similarly, the CMHA (2026) has cautioned that general purpose chatbots are not substitutes for qualified mental health services and may be harmful in crisis contexts. CMHA cites evidence that these systems reproduce societal biases and display elevated stigma toward substance use disorders and severe and persistent mental illness.
In summary, the current concerns centre around the increasing reliance on AI for mental health support in our digitally connected world, particularly impacting vulnerable populations like youth whose cognitive development is still ongoing. This could have lasting effects on their perception and application of empathy, potentially creating a disconnect between their cognitive and emotional processing. As professionals, we must remain vigilant about the potential negative effects of unregulated AI support on youth mental health.
The rapid integration of AI into our daily lives is unprecedented, surpassing our previous experiences. We want to keep the conversation open. As professionals, regardless of your field, how has your experience with AI, particularly chatbots like ChatGPT, affected your interactions with others? Have you noticed any evident changes in cognitive processing, genuine human connection, and the experience of empathy that cannot be replicated through generated conversations?
None.
The authors have no conflicts of interest to declare.
None.
∗Faculty of Behavioural Sciences, Yorkville University, Fredericton, NB, Canada.
Alawida, M., Mejri, S., Mehmood, A., Chikhaoui, B., & Isaac Abiodun, O. (2023). A comprehensive study of ChatGPT: Advancements, limitations, and ethical considerations in natural language processing and cybersecurity. Information, 14(8), 462. https://doi.org/10.3390/info14080462 [CrossRef]
Balan, R., & Gumpel, T. P. (2025). ChatGPT clinical use in mental health care: Scoping review of empirical evidence. JMIR Mental Health, 12, e81204. https://doi.org/10.2196/81204 [CrossRef] [PubMed]
Canadian Institute for Health Information. (2025). Mental health and substance use services, 2025. https://www.cihi.ca/en/taking-the-pulse-measuring-shared-priorities-for-canadian-health-care-2025/mental-health-and-substance-use-services-2025
Canadian Mental Health Association. (2026). More people in Canada are using AI as a mental health care tool, but are we ready for it? https://cmha.ca/news/ai-mental-health/
Ding, X., & Barbic, S. (2025). Perception of AI use in youth mental health services: Qualitative study. Journal of Participatory Medicine, 17, e69449. https://doi.org/10.2196/69449 [CrossRef] [PubMed]
Giansanti, D., & Cosenza, C. (2026). Artificial intelligence and youth: Cognitive, educational, and behavioral impacts. AI, 7(4), 121. https://doi.org/10.3390/ai7040121 [CrossRef]
Iftikhar, Z., Xiao, A., Ransom, S., Huang, J., & Suresh, H. (2025). How LLM counselors violate ethical standards in mental health practice: A practitioner-informed framework. Proceedings of the AAAI/ACM Conference on AI, Ethics, and Society, 8(2), 1311–1323. https://doi.org/10.1609/aies.v8i2.36632 [CrossRef]
Jacob, L., & Reddy, K. J. (2024). Role of cognitive and emotional empathy in the development of social cognition among adolescents. IP Indian Journal of Anatomy and Surgery of Head, Neck and Brain, 10, 4–9. https://doi.org/10.18231/j.ijashnb.2024.002 [CrossRef]
Kourgiantakis, T., Markoulakis, R., Lee, E., Hussain, A., Lau, C., Ashcroft, R., Goldstein, A. L., Kodeeswaran, S., Williams, C. C., & Levitt, A. (2023). Access to mental health and addiction services for youth and their families in Ontario: Perspectives of parents, youth, and service providers. International Journal of Mental Health Systems, 17(1), 4. https://doi.org/10.1186/s13033-023-00572-z [CrossRef] [PubMed]
KPMG Canada. (2025, October). Generative AI boom among Canadian students raises dilemmas [Press release]. https://kpmg.com/ca/en/media/2025/10/generative-ai-boom-among-canadian-students-raises-dilemmas.html
Mitchell, R. H., & Ameis, S. H. (2025). Strengthening mental health services for children and youth in Canada. The Canadian Journal of Psychiatry, 70(5), 341–346. https://doi.org/10.1177/07067437251346090 [CrossRef] [PubMed]
Moore, J., Grabb, D., Agnew, W., Klyman, K., Chancellor, S., Ong, D. C., & Haber, N. (2025). Expressing stigma and inappropriate responses prevents LLMs from safely replacing mental health providers. In Proceedings of the 2025 ACM Conference on Fairness, Accountability, and Transparency (pp. 599–627). Association for Computing Machinery. https://doi.org/10.1145/3715275.3732039 [CrossRef]
Nazir, T. (2026). ChatGPT for mental health support: A systematic scoping review of human–computer interaction implications. Health Education & Behavior, 10901981261435480. https://doi.org/10.1177/10901981261435479 [CrossRef] [PubMed]
SickKids Child Health Policy Accelerator. (2026). AI chatbots and companions: Implications for children and youth [Brief submitted to the Standing Senate Committee on Social Affairs, Science and Technology]. https://www.sickkids.ca/siteassets/about/child-health-policy-accelerator/ai-chatbots-and-companions.pdf
The Canadian Press. (2026, April 30). Young Canadians want AI companies to make their chatbots less addictive: Report. CP24. https://www.cp24.com/news/canada/2026/04/30/young-canadians-want-ai-companies-to-make-their-chatbots-less-addictive-report/
*Correspondence to: Julia Bertucci, Yorkville University, Yorkville Landing, Suite 102, 100 Woodside Lane, Fredericton, NB E3C 2R9, Canada. E-mail: julia.bertucci@yorkvilleu.ca
This work is distributed under the Creative Commons BY-NC-ND license. To view a copy of this license, visit http://creativecommons.org/licenses/by-nc-nd/4.0/. For commercial re-use, please contact sales@sgpublishing.ca.
Journal of CSWB, VOLUME 11, NUMBER 3, September 2026