Principle 1: Good governance builds on existing practices and adapts to meet a shifting AI landscape

Healthcare organizations already manage clinical changes, EHR workflows, and data assets with existing IT infrastructure. AI governance should apply the same foundational principles for the safe, compliant, ethical use of technology and align with existing processes.

Developing AI governance also requires clear roles, defined decision domains, and a consistent intake and evaluation workflow that aligns with existing quality, safety, and ethics committees.

That structured approach — which includes assessing current state, designing the governance model, and piloting with real use cases — helps ensure the governance process is practical, repeatable, and able to evolve as AI capabilities mature.

Additionally, traditional processes for developing governance models can be too slow for AI’s pace. Organizations need approaches that adapt in manageable, structured increments and have enough built-in flexibility to account for an uncertain and rapidly changing regulatory landscape.

Ultimately, effective AI governance is both adaptive and builds on existing structures to ensure each step forward matches your organization’s current capabilities, risk tolerance, and strategic priorities. By extending familiar governance structures and updating them with more adaptive processes, organizations can create a model that is both grounded and equipped to evolve as technology and regulations shift.

Principle 2: Strong data governance is a prerequisite for effective AI

AI models learn from the data they’re trained on and require large volumes of accurate, consistent, and well-structured information. Strong data governance ensures data is high-quality, accessible, and used appropriately, so models aren’t learning from incomplete, outdated, or biased information.

Good AI governance reinforces and elevates these data governance practices. As organizations adopt more advanced AI tools, the level of transparency, documentation, and oversight increases. Requirements such as model explainability demand stronger data documentation, clearer data definitions, more robust metadata, and better monitoring of how data flows into and through AI systems.

In this way, AI governance and data governance reinforce each other, creating a cycle where well-governed AI leads to better data practices, and better data practices help enable more trustworthy and effective AI.

Principle 3: Implementing governance requires organizational readiness

Most organizations underestimate the effort needed to introduce AI to clinical and operational staff. Anxiety about AI’s effect on jobs, a limited understanding of how AI works, and misaligned goals across organizations or service lines can all undermine effective implementation. These cultural dynamics matter more now as AI tools move closer to the point of care and into
day-to-day workflows.

Cultural readiness includes the operational and educational groundwork an organization must complete before AI governance structures can effectively function. As a core element of change management, readiness work prepares people and processes for the shifts AI will introduce. It includes transparent communication about how AI will be used in an organization, education to build AI fluency, and clear intake and evaluation processes for AI tools.

Principle 4: Maturity models help organizations progress responsibly

Maturity models for AI give health systems a structured way to assess their current capabilities and chart a realistic path forward. Rather than treating AI as a single, monolithic goal, these models break progress into stages. This helps healthcare leaders understand where they stand and determine what they need to progress.

During the early stages of maturity, organizations typically build foundational awareness and experiment with contained, low-risk use cases. They pilot AI tools in a single functional area to address clear pain points, such as using ambient listening to reduce the time clinicians spend on documentation. These early efforts help organizations test governance processes, validate value, and build familiarity with AI.

As systems progress into higher maturity levels, AI adoption becomes more coordinated and strategic and extends into broader workflows such as real-time claims processing or advanced outreach models. At higher levels, organizations also strengthen data governance, technical infrastructure, and cross-functional oversight.

At the most advanced levels, AI is embedded into the organization’s operating model and is integral to care coordination, operational optimization, and financial planning.

Across all stages, maturity models help health systems progress responsibly, ensuring adoption is anchored in governance, data quality, and organizational strategy rather than speed alone.

Final thoughts

As health systems navigate rapid AI advancement and shifting regulatory expectations, governance will remain the anchor that keeps innovation safe, strategic, and sustainable. This work is an ongoing commitment that strengthens data foundations, aligns stakeholders, and equips organizations to adopt AI responsibly. By continuously approaching governance with both rigor and agility, healthcare leaders can ensure that AI delivers meaningful, lasting value for clinicians, patients, and the broader organization.