AI Scribes Are Coming to a Chart Near You
What CDI Specialists Need to Know
Ambient AI tools that listen to a patient encounter and draft the clinical note in real time are already in use across Canadian hospitals and primary care clinics. Adoption is moving faster than the governance around it. In the past several months, a national funding program has scaled up, a large provincial trial has published its design, and researchers and privacy commissioners have started pushing back on some of the claims. CDI Specialists now have a fuller picture than we had even a year ago.
Why This Matters for CDI
If AI tools are drafting a growing share of the clinical narrative, that has direct implications for CDI work. It affects what we review, and how we frame queries. It raises a newer question CDI hasn't had to ask before: whether the documentation in front of us reflects what a clinician actually observed, or what a model inferred from an audio recording.
The BC Scribe Trial
A summary in Canadian Healthcare Technology in January 2026 describes the BC Scribe Trial - a coordinated, multi-site evaluation led by the Provincial Health Services Authority (PHSA). It's believed to be Canada's largest coordinated evaluation of AI scribe technology at the hospital and health-authority level.1 The trial was designed to evaluate use across approximately 8,000 providers, and rather than issuing a traditional RFP for a single vendor, PHSA brought six AI scribe tools already in informal use into a structured evaluation against Canada Health Infoway's Technology Evaluation Framework.1
A National Program, With Early Numbers Attached
Separately, Canada Health Infoway launched a national AI Scribe Program in June 2025, funding one-year licenses for pre-qualified AI scribe tools for primary care clinicians across the country.2 More than 12,000 clinicians have enrolled, and Infoway's own reporting on early results includes some specific figures worth flagging:
• More than 80% of participating clinicians reported feeling more engaged during patient visits, and about 60% perceived an improvement in the quality of care they provided.3
• More than 90% agreed the tool added value to their practice, with an average of roughly one additional hour saved per week during work hours.3
• A pre-implementation baseline survey of 2,658 primary care providers found 83.3% described their administrative workload as tiring, and 92.1% said administrative tasks reduce the time available for patient care.3
These are promising early figures, from the program's own evaluation reporting and not peer reviewed.
What the Peer-Reviewed Research Shows
A 2025 Canadian systematic review in the international journal Healthcare looked at the existing published research on AI scribes, and the picture is more cautious than most program announcements or vendor marketing.⁴ The review cites Canadian Medical Association data suggesting that about 60% of physicians see administrative burden as a direct contributor to declining mental health, and that Canadian physicians spend an estimated 18.5 million hours a year on unnecessary administrative tasks.⁴ Against that backdrop, the reviewed studies suggest AI scribes can genuinely reduce documentation burden, but the authors note that the evidence overall remains limited and inconsistent across studies, that accuracy varies significantly by tool and implementation, and that human review remains necessary, particularly for complex cases.⁴
The Case for Caution: Privacy, Accuracy, and Equity
An April 2026 piece in Policy Options lays out the risks that tend to get less attention in program rollouts. ⁵ Canada does not currently have a comprehensive federal AI regulatory framework for health care, since Bill C-27 died before passing; in its place, the BC and Ontario privacy commissioners have each issued their own guidance for health organizations using AI scribes.⁵ The piece raises several concerns directly relevant to documentation integrity:
• AI scribes can prioritize conversation flow over clinical nuance, which may lead to omission of details mentioned only briefly, and transcription errors can arise from accents, background noise, or multiple speakers.5
• Like other generative AI tools, scribes can produce output that sounds plausible but is factually incorrect: the same hallucination risk seen elsewhere in generative AI.5
• BC's privacy commissioner has specifically warned that health organizations should not rely on implicit consent, and that consent should be obtained in writing rather than assumed.5
• Some of the research cited in the piece has found documentation gaps between what's said in a consultation and what's transcribed, with higher error rates reported for some patient populations than others.5
None of this means AI scribes shouldn't be used. It means the accuracy and completeness of AI-drafted documentation can't simply be assumed, which is exactly the kind of scrutiny CDI Specialists already apply to physician-authored notes.
A Canadian Angle, and What Comes Next
Most of what's written about AI scribes online is American and tends to focus on billing optimization. The Canadian conversation, so far, is centered more on governance: provincial privacy commissioner guidance, EMR integration, and public-sector procurement, alongside a national funding program moving faster than the regulatory framework around it.
More provinces may publish their own guidance, CDI and health information management teams may get pulled further into decisions about what counts as acceptable AI-drafted documentation, while research continues to catch up with how quickly these tools are being deployed.
A Few Questions Worth Asking
As the AI Scribe conversation continues, it's worth asking a few questions: who reviews AI-drafted notes for clinical accuracy? Does the query process account for how AI-generated language differs from a physician's own phrasing? Is written, explicit patient consent being obtained, and is CDI represented in the conversation about how these tools are governed?
If your organization is already piloting or using an AI scribe, ACDIS Canada would like to hear about your experience at acdiscanadaleadership@gmail.com.
Drafted with AI assistance and reviewed and edited by ACDIS Canada
Sources
1. Arnaout, A., Besworth, M., & Brooks, N. (2026, January 30). BC conducts large-scale trial of six different AI scribes for care providers. Canadian Healthcare Technology. https://www.canhealth.com/2026/01/30/bc-conducts-large-scale-trial-of-six-different-ai-scribes-for-care-providers/
2. Canada Health Infoway. (2025). Evaluating Artificial Intelligence (AI) Scribes to Reduce Administrative Burden and Enhance Data Interoperability in Canadian Primary Care. https://insights.infoway-inforoute.ca
3. Kaminski, J. (2025). AI Scribes in Canadian health care: Easing documentation burden, enhancing care. Canadian Journal of Nursing Informatics, 20(1). https://cjni.net/journal/?p=14276
4. Sasseville, M., et al. (2025). The impact of AI scribes on streamlining clinical documentation: A systematic review. Healthcare, 13(12), 1447. https://doi.org/10.3390/healthcare13121447
5. Beny, H. (2026, April 1). AI scribes in health care raise risks for patients and privacy. Policy Options. https://policyoptions.irpp.org/2026/04/ai-scribes-health-care-canada-privacy-safety-risks/
Comments