CARF implements first AI standard in health and human services accreditation

CARF International is the first accreditor to implement an artificial or augmented intelligence (AI) standard requiring written policies and procedures if AI is used in relation to the delivery of a program/service.

“It is apparent that AI is going to occupy a significant role in most businesses. As adoption in healthcare is still in the early stages, it is important for CARF to set the bar that ensures the best possible outcomes for persons served,” said Michael Johnson, Senior Managing Director of Behavioral Health, CARF International. “Thoughtful implementation of AI requires organizations to be transparent and aware of the downside risks to persons served. Creating sound governance structures to analyze results and protect from bias is essential.”

AI standard rationale

Organizations seeking accreditation must demonstrate human oversight and accountability in several ways. Human oversight is the final decision-making authority in the continuous monitoring and auditing of AI outputs.

Accountability measures include regular training and education for those using and performing oversight of the AI tool(s); establishing a governance committee to assess vendor and data transparency; and ensure the AI tool(s) used comply with ethical and legal requirements.

“AI relies on data to function, and if data of persons served is used, there could be risk to privacy and data security,” said Terrence Carolan, Managing Director of Medical Rehabilitation and Aging Services, CARF International. “Having an AI standard that highlights disclosure to persons served is so impactful and consistent with CARF’s approach to accountability and transparency of accredited programs.”

AI standard elements

The standard addresses the essential components and elements of what AI policies and written procedures should include:

• How the organization uses AI
• Disclosures to the persons served regarding the use of AI in service delivery and for health data analysis and processing
• Privacy of health information and protection of sensitive data
• Human oversight and accountability
• What constitutes an incident requiring a response within the organization and to the persons served
• AI policies are reviewed at least annually and updated as needed

CARF standard in effect

The standard is in CARF’s ASPIRE to Excellence^® quality framework which is in all CARF standards manuals and will go into effect on July 1, 2026.

“The CARF AI standard establishes clear guidance around ethical use, transparency, and risk management while remaining practical and adaptable for organizations at varying levels of AI usage. Additionally, the standard supports responsible implementation without being overly prescriptive,” said Theresa Lindberg, Managing Director of Child and Youth Services, CARF International.