Skip to main content

What is AI policy?

The AI Glossary

In our classes, we often pause to explain the ideas behind AI and the language used to describe it. These posts grew out of those conversations. They’re for anyone who would like to understand AI better, with no technical background assumed.

AI policy is a set of rules, commitments or priorities intended to shape how AI is developed and used. A policy can belong to a government, an institution or a professional body. Its authority depends on who issued it and how it is implemented.

The word is broader than regulation. A law or binding regulation creates legal obligations. An organizational policy can set internal requirements, such as which tools staff may use with confidential information. A strategy describes priorities and choices, while governance assigns responsibilities for making and enforcing decisions. These ideas overlap but are not interchangeable.

A useful policy says more than “use AI responsibly.” It identifies the activity covered, the people responsible, the permitted and prohibited uses, and what evidence or approval is needed. It also explains what happens when the rules conflict with an urgent practical need or when a problem is discovered.

Policies make trade-offs visible. Restricting a tool may reduce one risk while creating another, such as pushing staff toward unapproved alternatives or leaving a harmful manual process unchanged. Encouraging adoption without adequate support can also create problems. The comparison should include realistic alternatives, not an idealized world without risk.

Public policy can influence access to data, research funding, competition, workforce development and accountability. Those choices shape which organizations can build or evaluate AI, not simply whether a particular model is technically capable.

In health care, consider a policy for AI-generated patient messages. It should distinguish drafting from sending, define when review is required, and preserve a route for patients who need a person. It should also address information access and record retention. The organization needs to assess whether the policy works in actual practice: can staff follow it during a busy day, and do patients understand the resulting service? A policy earns its value through the behavior and outcomes it produces, not the confidence of its wording.

Sources: WHO — Ethics and Governance of Artificial Intelligence for Health (2021); NIST — Artificial Intelligence Risk Management Framework (AI RMF 1.0, 2023)

To learn more, take a look at the AI in Health Care Certificate of Specialization from the Advanced Learning Academy at Harvard T.H. Chan School of Public Health. You can take an individual course or work toward the full certificate.


AI is a fast-moving field. This explanation reflects our class discussions and related material at the time of writing. Terminology, capabilities and tools may change.

For suggested additions or corrections, please email me at tpanch@hsph.harvard.edu.

Dr. Trishan Panch
Program Director, AI in Health Care
Harvard T.H. Chan School of Public Health — Advanced Learning Academy

Lead with AI in Health Care

The AI Pragmatist is a monthly newsletter from ALA faculty, with new explainers & essays from the Advanced Learning Academy