URAC Awards First Health Care Artificial Intelligence Accreditations

The first cohort spans three distinct health care operating contexts:

• •Guidehealth received accreditation as both an AI developer and user for AI-enabled value-based care and population-health solutions. Healthcare Innovation reports that its tools help health plans and health systems connect patients to care programs and support nurses' care-management work.
• •RediMinds received accreditation as both a developer and user for technology supporting its Center for Independent Dispute Resolution. URAC states that the company uses AI for data workflows and case-management optimization in that center.
• •SandsRx, a specialty pharmacy, received accreditation for AI used to streamline pharmacy operations and expand patient services, according to URAC. Healthcare Innovation reports that the evaluation covered safeguards for patient safety and privacy.

"As artificial intelligence plays a larger role in health care, strong governance helps organizations deploy these technologies safely, ethically and transparently," URAC President and CEO Shawn Griffin said in the organization's announcement. "Guidehealth, RediMinds and SandsRx are setting an important example for the industry, and we congratulate them for advancing innovation with accountability."

Guidehealth Chief Growth and Technology Officer Michael Gleeson described the accreditation as evidence that its approach emphasizes oversight, patient safety, and responsible implementation, according to Healthcare Innovation.

HIT Consultant reports that URAC launched its AI accreditation framework in 2025. BenefitsPro characterizes the process as voluntary and privately operated, and reports that the standards include transparency, testing, bias management, and data-security provisions. The program is open to organizations developing health AI systems and to organizations using such systems to provide services, according to BenefitsPro.

The distinction between developer and user accreditation is material for technical teams. A model developer can maintain documentation, evaluation artifacts, and version controls, but health systems and service providers also need operational controls around deployment, data access, monitoring, escalation, and human review. Healthcare Innovation reported in 2025 that Griffin cited guardrails, liability, oversight, and patient consent as areas the standards-development effort was examining.

HIT Consultant reports that the RediMinds evaluation led to more formal policies for unvetted employee AI use, restrictions on data entered into public large language models, and expanded transparency disclosures. Those reported controls illustrate issues that occur across enterprise AI deployments, particularly where employees can access general-purpose models outside approved data and identity boundaries.

For ML and data-governance practitioners, the announcement adds a health care-specific third-party assurance mechanism alongside internal model-risk processes and applicable regulatory obligations. Voluntary accreditation does not itself establish that a model is clinically effective or appropriate for every use case. However, industry adoption of comparable programs can make auditable governance evidence, such as documented monitoring, risk assessments, data controls, and human-in-the-loop procedures, more relevant in procurement and vendor-review workflows.