The AMA wants every health AI agent to answer four questions

An AI agent that finishes a multi-step clinical task without a human watching can do real harm before anyone notices. That is the problem the American Medical Association is trying to address with a new governance framework for agentic AI, published on Oct. 15, 2026.

The framework rests on one principle. If an agentic tool performs a clinical function, it must be backed by evidence and keep up high-quality performance. The AMA says it will use the document to guide its work with other organizations on safe, responsible use of agents in tasks that touch patient care.

How does it work?

The framework is built around four questions the AMA will ask when assessing any clinical AI agent:

• What clinical function is the AI performing?
• What shows it can perform that function, and what happens if it fails?
• What should the AI be permitted to do, and under what conditions?
• Who is responsible for the risks and outcomes?

These are not technical specifications. They are a way of forcing clarity about scope, proof, limits and liability. The framework states that "evidence, authority, oversight and accountability must remain aligned in service of better patient care." Agentic AI differs from generative AI because it can act on its own across several steps. Current uses include ambient documentation, diagnostics and patient monitoring.

Why does it matter?

Adoption is moving fast. In a recent survey of 250 U.S. healthcare leaders, about 28% said their organization has already deployed agentic AI, and 44% were running pilots or proof-of-concept projects. A majority expect a transformative effect on clinical workflows.

But oversight is lagging. Some 72% of respondents said agentic tools are occasionally put in place without IT approval. More than half named compliance or regulatory violations and unauthorized access to protected health information as major concerns. So the technology is arriving faster than the controls around it.

For hospital executives in the GCC, that gap should sound familiar. Many health systems in the region are scaling digital tools quickly under national strategies. Shadow deployments and unclear accountability are risks any large provider can face, whatever the country.

The context

The United States has no substantive federal AI regulation for healthcare. Industry groups have stepped in with guidance and guardrails instead. States have produced a patchwork, with some restricting the technology and others experimenting with AI-based care. According to the firm Manatt, 47 states introduced more than 250 AI bills affecting healthcare in 2025.

The AMA is also reasserting the physician's central role in building and deploying these tools. That position is clear, though a framework of questions is not a binding rule.

The Gulf has a chance to learn from this. Saudi Arabia under Vision 2030, the UAE through its national health and AI strategies, and other GCC states are investing heavily in digital health. Regulators and providers here can write accountability into procurement and licensing early, rather than retrofitting it later. The AMA's four questions are a practical starting point: ask them of every vendor, before an agent touches a patient record.