AIHealthcare Analytics

HAARF — Healthcare AI Agents Regulatory Framework

The Healthcare AI Agents Regulatory Framework (HAARF), published as a preprint on medRxiv, proposes a comprehensive security verification standard for autonomous AI systems operating in clinical environments. As agentic AI moves from concept to deployment in health systems, HAARF represents an early attempt to define what regulatory and security verification should look like for AI that takes autonomous actions—not merely provides recommendations.

Framework Scope. HAARF covers the verification of autonomous AI systems across dimensions including safety, security, reliability, and accountability. It is designed specifically for clinical environments where autonomous AI actions could directly affect patient safety—distinguishing it from general enterprise AI governance frameworks that may not account for the unique risk profile of clinical deployment.

Relationship to Existing Regulation. HAARF is positioned as complementary to FDA SaMD regulation and existing clinical AI governance standards, addressing a gap: current regulatory frameworks were designed primarily for AI systems that support human decision-making, not for systems that act autonomously. As Amazon Connect Health, agentic platforms from Epic, and other autonomous clinical AI systems proliferate, the need for frameworks like HAARF becomes more acute.