Summary
AI arrived in healthcare faster than most security and compliance programs could respond. Ambient scribes, clinical decision support, claims automation, and AI features embedded in vendor platforms are already in use. Many organizations are now trying to govern tools that are already in production.
AI governance in healthcare works best when it is built into your existing governance, risk, and compliance (GRC) program rather than run as a separate initiative. Governance sets decision rights and acceptable use. Risk management inventories and assesses AI use cases. Compliance maps AI controls to frameworks like the National Institute of Standards and Technology (NIST) AI Risk Management Framework and ISO/IEC 42001. Third-party risk management extends oversight to the AI inside vendor products.
This article explains why AI governance can’t wait, where each element fits inside a healthcare GRC framework, and how Meditology Services helps providers, health plans, and health tech companies govern AI without slowing adoption.
Why AI Governance in Healthcare Can’t Wait
The most immediate AI risk for most healthcare organizations isn’t an outside attacker. It’s shadow AI: staff using consumer AI tools that the organization hasn’t approved. A clinician pastes a clinical summary into a public chatbot to draft a patient letter. A billing coordinator uploads claims data to speed up coding. Each case creates an unmonitored path for protected health information to leave the organization.
The data reflects how wide the gap is:
- IBM’s 2026 Cost of a Data Breach Report found that more than 20 percent of organizations reported a breach targeting AI models or applications.
- A Wolters Kluwer survey found that only 27 percent of physicians and nurses were aware of formal AI governance policies in their organizations.
- A Premier Inc. analysis found that 80 percent of health systems lack internal governance standards to guide AI adoption.
In interviews for Meditology’s 2026 Healthcare Security Outlook, chief information security officers (CISOs) and senior security leaders consistently delivered the same message: governance must come before enablement. The question is where that governance should live.
Where AI Governance Fits in Your GRC Framework
Many organizations treat AI governance as a new program with its own committee, policies, and tracking. That approach creates duplication and gaps. A more sustainable model places AI governance inside the GRC structure you already operate, extending each component to cover AI. Some teams call this AI GRC: the same operating model, applied to a new class of technology risk.
The table below shows how AI governance maps to the four pillars of a healthcare GRC operating model.
| GRC Pillar | AI Governance Components | Existing Process to Extend |
|---|---|---|
| Governance | AI governance committee, charter, decision rights, AI acceptable use policy | Security and compliance committees, policy management |
| Risk | AI use case inventory, risk tiering, AI risk assessments, risk register entries | Enterprise and security risk assessments, risk register |
| Compliance | Control mapping to NIST AI RMF, ISO/IEC 42001, the Health Insurance Portability and Accountability Act (HIPAA), and emerging state laws | Control library, framework mapping, audit readiness |
| Ecosystem Risk | Vendor AI disclosure, AI questions in vendor assessments, contract terms | Third-party risk management (TPRM) program |
Governance: Decision Rights and Acceptable Use
AI governance starts with deciding who approves AI use and on what basis. The most mature healthcare organizations use a cross-functional AI governance committee that brings together clinical, legal, compliance, privacy, and security leadership. That committee evaluates AI tools before they reach clinical or administrative settings.
An AI acceptable use policy defines which tools are approved, what data can be used, and what is prohibited. Policy alone isn’t enough. Effective organizations pair the policy with technical controls, such as data loss prevention tuned to detect AI-bound data flows, and with approved alternatives that give staff a sanctioned way to get the efficiency they’re looking for.
Risk: Inventory, Tiering, and Assessment
You can’t govern what you can’t see. An AI use case inventory should capture internally built models, AI features in purchased software, and tools staff are already using. From there, tier each use case by its potential impact. A tool that drafts internal emails carries different risk than one that influences clinical decisions or member coverage.
Higher-risk use cases need structured AI risk assessments. These should evaluate data handling, model transparency, bias, security of the deployment environment, and patient safety implications. Findings belong in the same risk register leadership already reviews, so AI risk is visible alongside other enterprise risks.
Compliance: Mapping to Frameworks You Can Defend
AI-specific frameworks give structure to controls and evidence. Rather than building a separate compliance track, map AI requirements into your existing control library and audit processes. This keeps evidence collection consistent and makes it easier to show regulators, customers, and auditors how AI is governed.
The regulatory environment is also moving quickly. According to Manatt Health’s AI Policy Tracker, state lawmakers have introduced more than 280 health AI bills so far in 2026, and 40 have been enacted across 24 states. Mapping AI controls into your GRC framework now makes it easier to adapt as requirements change.
Ecosystem Risk: The AI Inside Your Vendors
Electronic health record (EHR) platforms, clinical decision support tools, revenue cycle systems, and other vendor products increasingly include AI features, sometimes without explicit disclosure. As one CISO put it, a governance discussion around AI policy raised an immediate follow-on question: what are the AI policies and practices of all our third parties?
Vendor AI use should become a standard part of vendor risk assessment. Ask what data is used to train or run models, whether outputs influence clinical or coverage decisions, and how model behavior is monitored. Extend contract language to cover AI disclosure and data use.
Choosing an AI Governance Framework: NIST AI RMF and ISO/IEC 42001
Two frameworks anchor most healthcare AI governance programs. Choosing an AI governance framework doesn’t require you to rebuild GRC from scratch.
- NIST AI Risk Management Framework (AI RMF): A voluntary framework from NIST organized around four functions: Govern, Map, Measure, and Manage. It aligns naturally with existing risk management practices and gives teams a common vocabulary for AI risk.
- ISO/IEC 42001: An international standard for AI management systems. It defines requirements for establishing, maintaining, and improving how an organization governs AI, and it can be certified by an accredited body.
HITRUST also offers an AI Security Assessment and Certification for organizations that deploy or provide AI systems. For health tech companies already pursuing HITRUST, this can extend existing assurance work to AI.
What AI Governance Looks Like Across Healthcare Organizations
The core model is the same across the industry, but priorities differ by organization type.
Health Systems and Providers
Clinical AI and shadow AI are the main concerns. Clinical leaders need a seat on the governance committee. Use cases that influence diagnosis, treatment, or documentation require patient safety review, and staff need approved tools that reduce documentation burden so they don’t turn to consumer alternatives.
Health Plans
AI often supports claims processing, utilization management, member services, and fraud detection. Governance should focus on how AI-supported decisions affecting members are reviewed, documented, and explained, along with oversight of the vendors and delegated entities that use AI on the plan’s behalf.
Health Tech Companies
For vendors, AI governance is both an internal control issue and a sales issue. Healthcare customers are adding AI questions to their security reviews. A documented AI governance program, mapped to recognized frameworks, helps answer those questions quickly and supports assurance options such as ISO/IEC 42001 or HITRUST AI certification.
How to Start Building AI Governance into Your GRC Program
AI governance matures in stages, from foundational structure to operational oversight to measurable, board-ready reporting. These steps reflect a practical sequence for most healthcare organizations:
- Charter an AI governance committee with clear decision rights, membership, and escalation paths.
- Build an AI use case inventory that includes internal tools, vendor AI features, and shadow AI.
- Publish an AI acceptable use policy backed by technical controls and approved alternatives.
- Add AI to your risk assessment process and record AI risks in the enterprise risk register.
- Extend TPRM to vendor AI through updated assessment questions and contract terms.
- Map AI controls to recognized frameworks and report AI risk to leadership alongside other enterprise risks.
AI Governance Readiness Checklist
Use these questions to see where your organization stands today:
- Is there a cross-functional committee with authority to approve or reject AI use cases?
- Do you have a current inventory of AI tools, including vendor AI features?
- Is there an AI acceptable use policy that staff know about?
- Are technical controls in place to limit patient data flowing to unapproved AI tools?
- Are higher-risk AI use cases assessed before deployment?
- Do AI risks appear in your enterprise risk register?
- Do vendor assessments ask how vendors use AI and data?
- Are AI controls mapped to a framework such as the NIST AI RMF or ISO/IEC 42001?
- Does leadership receive regular reporting on AI risk?
Frequently Asked Questions (FAQ)
What is AI governance in healthcare?
AI governance in healthcare is the set of policies, decision rights, risk processes, and controls that determine how an organization approves, uses, and monitors AI. It covers internally built tools, AI features in vendor products, and staff use of external AI services, with particular attention to patient data and patient safety.
Where does AI governance fit in a GRC program?
It fits inside each part of an existing GRC program. Governance defines who approves AI and how. Risk management inventories and assesses AI use cases. Compliance maps AI controls to frameworks. Third-party risk management extends oversight to vendor AI.
Who should sit on an AI governance committee?
Most healthcare organizations include clinical leadership, security, privacy, compliance, legal, IT, and data or analytics leaders. Health plans often add utilization management or member services leaders. The committee needs clear authority to approve, restrict, or reject AI use cases.
What is the NIST AI Risk Management Framework?
The NIST AI RMF is a voluntary framework that helps organizations manage AI risk. It is organized around four functions: Govern, Map, Measure, and Manage. Many healthcare organizations use it to structure AI risk assessments and align them with existing risk management practices.
Do we need ISO/IEC 42001 certification?
Not necessarily. ISO/IEC 42001 certification can be valuable for health tech companies that want to demonstrate mature AI governance to customers. Many provider and health plan organizations use it as a reference framework without pursuing certification.
What is shadow AI, and why does it matter in healthcare?
Shadow AI is the use of AI tools that an organization hasn’t approved, such as consumer chatbots. In healthcare, it creates an unmonitored path for protected health information to leave the organization. Approved alternatives, clear policy, and technical controls help reduce it.
About the Author
Alan DeVaughan | Director, IT Risk Management
Alan DeVaughan is an experienced compliance and information security director with more than a decade of expertise supporting organizations through SOC 2 readiness assessments and examinations. As the lead for Meditology’s SOC 2 service line, he also ser. ves as a consultant team leader, advising healthcare organizations of varying sizes and complexity on IT, privacy, security, and regulatory compliance. Alan brings deep knowledge of leading security and compliance frameworks, including NIST, HITRUST, SOC 1 and SOC 2, HIPAA, and FFIEC. With a background in network administration, he has over 25 years of experience in information technology consulting across a wide range of industries.