BC conducts large-scale trial of six different AI scribes for care providers

As generative artificial intelligence tools become more widely available, healthcare organizations across Canada are confronting a new operational reality: clinicians are increasingly adopting AI tools on their own initiative. This trend – often described as Bring Your Own AI (BYOAI) – reflects both unmet clinical needs and growing familiarity with consumer AI platforms (e.g. ChatGPT) that support summarization, drafting, and information synthesis.

While such tools can improve efficiency, their unregulated use in clinical contexts introduces significant risk to health authorities and organizations. Patient privacy, clinical data security, regulatory compliance, and professional liability remain unresolved when consumer-grade AI is used outside institutional controls. Guidance from the Canadian Medical Protective Association (CMPA) has reinforced that clinicians remain accountable for AI-assisted care, even as legal and regulatory frameworks continue to evolve.

Attempts to ban BYOAI have largely proven ineffective. Research from the MIT Center for Information Systems Research suggests that prohibitions tend to push AI use underground, increasing organizational exposure to privacy and security risks rather than reducing them. The reality is that AI is already embedded in clinical practice, whether health systems formally acknowledge it or not.

In British Columbia, that shift has led to the BC Scribe Trial – a coordinated, multi-site evaluation of ambient AI scribe technologies designed to provide a managed, enterprise-approved alternative to unmanaged BYOAI use.

A first-of-its-kind evaluation at scale: The BC Scribe Trial is believed to be Canada’s first multi-vendor, multi-site randomized evaluation of third-party non-enterprise integrated AI scribes within a publicly funded healthcare system. Led through the Provincial Health Services Authority (PHSA) in collaboration with partner health authorities, the program spans multiple electronic medical record (EMR) platforms and clinical environments.

The B.C. initiative builds on recent work by OntarioMD, which evaluated AI scribes primarily within family medicine and determined that these tools helped physicians’ daily work in reducing documentation burden and cognitive overload.

B.C.’s approach differs in scope. Rather than focusing solely on individual clinician experience, the BC Scribe Trial examines whether third party individual AI scribe solutions can be governed, supported, and scaled responsibly across a complex public healthcare system. The trial was designed to support and evaluate AI scribe use across approximately 8,000 providers, making it one of the largest coordinated AI initiatives undertaken in Canadian healthcare to date.

There was no traditional RFP by design. Instead of endorsing a single vendor, the program deliberately brought six AI scribe tools already in widespread, unregulated use into a structured, health authority-compliant environment. The focus was system readiness, not product selection: determining whether commonly used tools could meet public-sector requirements for privacy, security, data stewardship, and clinical governance.

Participating clinicians rotate across solutions, enabling direct comparison of performance, usability, workflow impact, and support requirements. This approach allows system leaders to assess how different tools function across specialties, encounter types, and care settings, while reducing vendor-specific bias.

The evaluation is guided by Canada Health Infoway’s Technology Evaluation Framework, ensuring that assessment extends beyond usability to include privacy, security, interoperability, clinical safety, and implementation readiness. Quantitative data, including usage metrics and surveys, are combined with qualitative feedback from clinicians to assess documentation quality, trust, and integration into real-world workflows.

From evaluation to provincial deployment: A distinguishing feature of the BC Scribe Trial is that it was designed from the outset with post-trial deployment in mind. Rather than concluding with a traditional pilot report, the program established pathways to enable broader access to approved tools across British Columbia.

Following the free-access trial, clinicians transitioned to a discounted self-pay model supported by a provincial master agreement. Participating vendors are offered the opportunity to enter negotiated, health authority–compliant contracts that meet standardized requirements for privacy, security, data handling, and clinical governance.

This structure enables deployment beyond early adopters, including expansion to the remaining B.C. health authorities, while preserving clinician choice and ensuring alignment with institutional standards – without requiring immediate centralized funding. By pre-negotiating contracts and standardizing requirements, PHSA has reduced one of the most common barriers to scale: the need for individual clinicians or departments to independently assess vendor compliance and risk.

Enterprise support and shared accountability with vendors: Scaling AI scribes across thousands of providers requires operational support beyond the technology itself. The BC Scribe Program leverages existing enterprise support structures, including health information management teams and education specialists who traditionally support front-end speech recognition and clinical dictation tools.

These teams provide onboarding, training, and workflow optimization, ensuring that third party AI scribe solutions are implemented consistently and safely. Vendors are assigned clearly defined roles and responsibilities, including technical support and issue resolution, with formal escalation paths between enterprise support and vendor customer support. This shared accountability model for technology support helps mitigate organizational risk in unmanaged BYOAI environments.

Enabling responsible market entry through a new private public partnership model: Beyond clinical evaluation, the BC Scribe Trial is also testing a new public–private partnership model for AI adoption in healthcare. AI companies have historically faced significant barriers to entering Canada’s publicly funded healthcare market. Through PHSA and the BC Scribe Program, B.C. has established a structured entry pathway that enables vendors to test, refine, and scale solutions within a controlled public-sector framework.

For vendors, the model provides access to clinical scale and diversity – across thousands of providers, multiple specialties, and several EMR environments – while aligning product development with public-sector requirements. The trial has become a “living lab” for vendors to test different communication, marketing, product design interventions, all while learning from each other in a competitive environment.

From BYOAI Risk to Opportunity: As generative AI continues to evolve, healthcare organizations face increasing pressure to distinguish between unmanaged consumer tools and institutionally supported AI solutions. The BC Scribe Trial offers a practical, governance-led response. By combining large-scale evaluation guided by a national framework, standardized contracts, enterprise support, and clear accountability structures, the program reframes BYOAI from a compliance risk into an opportunity for responsible system-wide adoption.

Dr. Angel Arnaout is a surgical oncologist, professor of surgery, scientist and chief medical informatics officer at Provincial Health Service Agency in BC. Dr. Maximillian Besworth (PhD) is an adjunct professor at the University of British Columbia and PHSA senior director of medical informatics. Naomi Brooks is the senior executive director for Lower Mainland Health Information.