Why HITRUST is a valuable framework for secure AI use in healthcare

AI is moving fast in healthcare. Tools for clinical documentation, revenue cycle management, coding and decision support are streamlining tasks across the industry. The efficiency gains are real, but they come with risks. The challenge is that many organizations don’t have a structured framework to evaluate or communicate how those risks are being managed.

HITRUST offers two frameworks to help healthcare organizations navigate the risks of AI adoption. With its well-recognized name in healthcare, the HITRUST AI options offer credibility with the people who matter most to your business.

Keep reading to learn how your organization can get started with this structured, independently validated path to demonstrate responsible use of AI.

What risks come with AI use in healthcare?

AI adoption in healthcare is improving efficiency and patient care while reducing burnout among clinical and administrative staff. These are obviously positives, but they introduce risks your organization needs to account for. Here’s where healthcare organizations are most exposed:

• Data flow and vendor depth: When using AI tools for billing and revenue cycle management, you need to be able to answer two questions: What kind of data are entering those systems? Where is the data going? The vendors you work with have vendors of their own, so tracking how far downstream your data travels and how it gets there is a real challenge.
• Output reliability: Can you trust what your AI tools are producing? Organizations are relying on AI outputs, but who’s validating accuracy? In healthcare, this affects patient care and data integrity. For example, a coding error driven by an unchecked AI model can cascade into billing inaccuracies, compliance exposure and payer disputes. These are problems that can be expensive and hard to unwind.
• HIPAA compliance: AI tools that process or retain protected health information (PHI) without proper controls can trigger HIPAA violations.
• AI quality and model integrity: AI technologies introduce considerations beyond traditional information security, including model integrity and reliability, data usage and training concerns, transparency and oversight and evolving regulatory expectations.

The benefits of AI are significant, but you need to have guardrails in place to mitigate the risk.

What are the HITRUST AI options?

HITRUST has expanded its framework to address AI-related risk by introducing AI-focused options. One option adds certifiable AI security controls to an existing validated assessment and the other provides non-certified AI risk management insights to support governance and transparency.

Understanding the difference between them is key to selecting the right option for your organization.

HITRUST AI Security Assessment and Certification (ai1/ai2)

The AI Security Assessment is designed for organizations that want certifiable, security-focused assurance over AI-enabled systems. This option can be applied to e1 or i1 assessments, resulting in an ai1 designation, or to r2 assessments, resulting in an ai2 designation.

A few important characteristics:

• The AI Security Assessment is not a standalone assessment. It’s an add-on to an existing validated assessment.
• It introduces a limited set of AI-specific security requirements — up to 44, depending on scope and tailoring.
• Certification outcomes depend on meeting both the underlying e1/i1/r2 requirements and the AI security scoring thresholds.

This path is best for organizations that want to show boards, partners, payers and patients that patient data is protected as AI use expands. HITRUST certification provides assurance that data is safe and that human leadership and expertise remain at the heart of your organization. It also helps separate businesses at the cutting edge of AI integration from those playing catch-up.

HITRUST AI Risk Management Assessment (non-certified)

HITRUST’s AI Risk Management option takes a different approach. Rather than focusing on certification, it emphasizes governance, oversight and life cycle risk management for AI. This option aligns with other frameworks, including the NIST AI Risk Management Framework (AI RMF) and ISO/IEC 23894:2023 for AI risk management.

The non-certified route provides guidance for assessing AI risk maturity, strengthening internal governance programs, supporting regulatory readiness and communicating responsible AI practices to stakeholders. Think of the two options this way: the Risk Management Assessment lays the foundation; the Security Certification proves to the people buying it that the house is structurally sound.

How to get started with HITRUST AI certification

The path forward depends on your starting point. Are you building a HITRUST program from scratch, or do you already hold a certification and want to add AI?

If you’re new to HITRUST

For businesses that don’t already hold HITRUST certification, the general path looks like this:

1. Choose your assessment level

Determine if e1, i1 or r2 is right for your business based on your current program maturity and what customers expect. Many organizations start with an e1 certification with the AI certification added on. Getting ready for an r2 can take one to two years. The e1 covers essential cybersecurity controls that are relatively straightforward to implement, so it doesn’t unnecessarily prolong the certification process.

2. Define your scope

Identify which AI-enabled systems are in scope and whether AI options apply.

3. Find an authorized assessor

A HITRUST-authorized external assessor is required for certification.

4. Complete a gap assessment

Understand where your current controls fall short of HITRUST AI requirements.

5. Design and implement controls

Develop the policies, procedures and technical controls needed to close the gaps.

6. Undergo a validated assessment

An independent assessor tests implementation and submits findings to HITRUST.

Whether you’re preparing for an e1, i1 or r2, adding AI considerations early in the scoping process is critical. Decisions about AI applicability impact assessment scope, evidence expectations, testing procedures and client and stakeholder communications.

If you already hold a HITRUST certification

Adding AI to an existing e1 or i1 is the most straightforward entry point into HITRUST AI certification. Adding the AI Security Assessment is done by selecting the “Security for AI Systems” compliance factor within MyCSF. When paired with an e1 or i1 assessment, it results in an ai1 designation.

A few practical things to know:

• The AI certification bolted onto an e1 or i1 results in a one-year certification. The same annual renewal cycle as the base e1 or i1.​
• For e1 and i1 assessments, you can carve vendors out of scope, which simplifies the process if your AI tools involve third-party providers. In this case, you’d only test the portion of the requirements that are your organization’s responsibility.
• HITRUST’s logic is that AI tools sit on top of an existing environment. The base certification addresses the environment’s security controls and the AI certification layer addresses the AI-specific tools on top of it.

Starting a readiness assessment early gives you time to identify and remediate gaps before the validated assessment, rather than rushing to close them afterward.

For organizations already certified at the r2 level, the most common approach is to start a readiness assessment and, when possible, align AI certification timing with the r2-validated assessment cycle. You can pursue the AI certification separately, creating a new object and inheriting controls from the existing r2, and then testing only the AI-specific controls. Over time, the two certifications can be synced into a single cycle.

Pairing the AI certification with an r2 results in a two-year certification, reflecting the additional rigor involved.

What makes HITRUST a good framework for AI risk management?

For healthcare organizations, there are several reasons HITRUST may be the ideal solution for AI risk management, including:

It was built for healthcare

Today, businesses across many industries seek HITRUST certification. But the framework was originally designed for healthcare and it is highly respected in the field. For provider organizations, HITRUST offers a recognized way to demonstrate to payers, health system partners, cyber insurers and boards how AI use is governed and secured.

It consolidates standards into one framework

HITRUST’s approach allows organizations to address AI-related considerations within existing assessment structures rather than creating a completely separate assurance model. It incorporates controls from frameworks such as ISO and NIST, creating a single, consolidated program rather than a patchwork of compliance activities.

It scales to your risk profile

Whether you’re working toward an e1 as a starting point or carrying an existing r2 certification, the decision to include AI options should be based on scope, maturity and assurance objectives — not simply the presence of AI tools. That flexibility matters. It means you’re not overburdened with controls you don’t need.

It requires third-party validation

HITRUST carries the credibility of independent validation. Healthcare organizations can demonstrate not only that they have controls in place but also that the quality of those controls has been externally verified. That distinction matters to stakeholders.

It quiets board-level anxiety about AI

For leadership teams and boards that have reservations about how AI is being used, HITRUST verifies that you have an appropriate framework in place and that your organization is ready to respond to increasing governance pressure.

Read more

• The outpatient behavioral health squeeze and the digital therapy opening
• AI in healthcare finance: Practicality over hype, strategy over speculation
• Revenue leakage in healthcare: How to find and prevent it