Blog
AI Governance in Healthcare: What Boards
Need to Ask Before Adopting Clinical AI
The hardest part of adopting clinical AI in a health system is rarely the technology — it's governance: who is accountable when the AI is wrong, where the data lives, and how the board can be confident the system stays safe after launch. This is a practical checklist for boards and executives evaluating a clinical AI deployment, drawn from building and deploying MedTalk AI inside a live public Digital Health Record, and from work on AI governance more broadly.
Why this is a board-level question, not just an IT one
Clinical AI adoption decisions increasingly sit with boards and executive committees, not just IT procurement — and for good reason. A scribe, a triage tool, or a clinical decision-support system that generates or surfaces patient-facing content carries clinical safety, privacy, and reputational implications that go well beyond a typical software purchase. I wrote about this shift in more detail in "The Industrialisation of AI" for Pulse+IT, and recently discussed AI governance frameworks directly with UK Minister Kanishka Narayan MP, the Parliamentary Under-Secretary for AI and Online Safety — the questions Australian health boards are asking now mirror what's being debated at a national policy level internationally.
The checklist
A starting point, not a finish line
None of this is meant to slow adoption down for its own sake — the deployment data from MedTalk AI's rollout inside Canberra Health Services shows real, measurable benefit when governance is built in from the start rather than retrofitted.
The boards and health systems that move fastest and most safely are the ones treating these questions as part of the procurement process from day one, not as a post-incident review.
If your board or health system is working through an AI governance framework or adoption policy, I'm happy to talk through what's worked and what hasn't —
Contact Us
