AI Governance in Health: What Ontario Can Learn from NSW

Picture a boardroom somewhere in Sydney, Australia. A taskforce of clinicians, administrators, and technology advisors is sitting around a table, and they are doing something most health systems have been avoiding for years. They are writing the rules.

Not after something goes wrong. Before it does.

New South Wales Health has launched a formal AI framework for its public hospital system, providing a consistent, governed path for adopting artificial intelligence across one of the largest public health systems in the Southern Hemisphere. And if you think that kind of structure is a long way from Ontario, I’d gently suggest you start paying attention.

Why a Framework Matters More Than the Technology Itself

I’ve said this before and I’ll keep saying it. AI is the guard dog. A guard dog in the calm, controlling hand of the handler is an incredible tool. Left to its own devices, on a long chain with no direction, it can cause serious harm.

The NSW framework gets that. It isn’t anti-AI. It’s the opposite. It’s a structure designed to let AI move faster, not slower, because every proposed project has a clear path to approval, a risk-based assessment process, and an expert advisory body to flag problems before they become incidents.

That is the handler. That is exactly what Ontario’s health system needs to be building right now.

At OpsMed, we take this seriously. We are not rogue operators pushing AI into clinical settings and hoping for the best. Every automation we build for Ontario family medicine clinics is reviewed by a human in the loop. That human, in our case, is me. Eighteen years as an Advanced Care Paramedic, chart auditor, and clinical reviewer. The AI does the heavy lifting. I check the work.

What a Risk-Based Approval Approach Actually Looks Like

The NSW model uses a risk-based approach to evaluate AI projects. Lower-risk tools can follow a more streamlined path. Higher-risk applications, anything touching clinical decision-making or patient safety, receive deeper scrutiny before deployment.

This is common sense dressed up in policy language, and I mean that as a compliment.

Think about what this looks like in a family medicine clinic in Hamilton or Niagara. An AI tool that helps an MOA (Medical Office Administrator, the person who manages your front desk, patient records, and about forty other things simultaneously) draft referral letters? Lower risk. Reviewable. Improvable. An AI tool that suggests a diagnosis or recommends a treatment pathway? That is a completely different conversation, and it should be treated as one.

OpsMed operates in that first category. We automate the administrative back office. Billing support, documentation, referrals, inbox management. We do not make clinical decisions. We never will. That line is not negotiable for us.

The NSW framework essentially draws that same line at a systems level. And that is where Ontario’s regulatory bodies, including the College of Physicians and Surgeons of Ontario and the Ontario Health Insurance Plan compliance structures, will eventually land too.

Ontario Isn’t There Yet, But the Tide Is Moving

We are at the pioneer stages right now. There is no formal Ontario AI health framework sitting on a government website with a risk-based assessment process and advisory panels. Not yet.

But the absence of a framework is not the same as the absence of risk. Clinics are already using AI tools, whether they call them that or not. Automated transcription. Smart scheduling. Predictive recall systems. The technology is in the building. The governance question is whether anyone is managing it with intention.

NSW decided to get ahead of it. That decision protects patients. It protects clinicians. And frankly, it protects the institutions from the kind of liability that comes when something goes wrong and there is no paper trail showing anyone thought carefully about the risks.

For Ontario family medicine, the lesson is this: the clinics that start thinking about AI governance now, that ask vendors the hard questions about compliance, data handling under PHIPA, and human oversight, will be far better positioned when formal frameworks arrive here. And they will arrive.

Either we get ahead of the technology tsunami, or we get crushed by it.

What We Are Doing About It at OpsMed

My brother Marc handles the technology architecture. My brother Steven brings MSSP-level security expertise through CyberLeda, our partner company for IT security and infrastructure. And I carry the clinical lens. Between the three of us, we cover the full stack: the automation, the security, and the clinical quality assurance.

We do not want to push boundaries. We want Ontario physicians and their teams to feel comfortable. That means transparent AI, documented workflows, and a human reviewer, me, making sure the output meets professional standards before it touches a patient record or an OHIP billing file.

When formal AI governance frameworks arrive in Ontario, and they are coming, we want our clinic partners to already be operating inside those lines. Not scrambling to catch up.

Good documentation is like music. You close your eyes and it brings you to the scene. The NSW framework is trying to make sure AI-assisted documentation stays that good, that trustworthy, that clear. We are trying to do the same thing for every clinic we work with in the Golden Horseshoe.

If you are a physician or clinic operator wondering whether now is the right time to think seriously about how AI fits into your practice, the answer NSW just gave the world is yes. And if you want to have that conversation with a team that comes from healthcare, not from a sales deck, we are ready when you are.

The absence of a framework is not the same as the absence of risk. Clinics are already using AI tools, whether they call them that or not.

Explore our clinic automation services or book a free check-up.