Healthcare leaders call for governance as agentic AI adoption grows

As healthcare organizations increasingly adopt agentic AI—tools capable of taking autonomous actions rather than just providing information—experts are calling for rigorous governance frameworks to manage clinical and administrative risks.

Research indicates significant momentum in the sector: 28% of surveyed healthcare organizations have agentic AI in production, 44% are piloting the technology, and 21% plan implementation within the next year. While these agents can reduce administrative burdens by scheduling appointments, managing patient flow, or handling prior authorizations, their autonomy introduces complexities regarding accountability.

Experts emphasize that healthcare leaders must establish clear written protocols defining who owns each AI agent, the scope of its autonomous authority, and the mechanisms for human oversight. There is a growing concern that if an agent makes an error, such as scheduling a patient with the wrong specialist or mismanaging clinical urgency, the legal and professional responsibility may fall on the delivery organization or the attending physician rather than the technology provider.

Recent developments highlight this shift toward active AI: ARPA-H has funded Kaiser Permanente to conduct a large randomized trial of agentic AI in live care, and OpenAI recently released a read-only ChatGPT connector for Epic. Meanwhile, a UK national commission has recommended staged authorization for AI-enabled devices to ensure continuous monitoring throughout their operational lifecycles.