How Many Medical Groups Have an AI Policy? 20% in 2026

In MGMA's January 2026 poll of 328 practice leaders, 20% of medical groups had a formal AI governance policy, and another 22% were writing one. The other 56% had no policy, and 2% were unsure.

What the January 2026 poll found

MGMA Stat ran the poll on January 20, 2026 and published the results on January 21. Of 328 applicable responses, 20% reported a formal AI policy in place and 22% reported one in development. That makes 42% with a policy or a plan.

Groups that have a policy most often review it once a year. MGMA lists quarterly, monthly and twice-yearly reviews as less common cycles.

MGMA also cited its joint research with Humana, run in June and July 2024 and published in November 2024. In that work, 73% of surveyed organizations had no formal governance structure for AI. The two figures come from different samples and different wording, so treat the drop from 73% to 56% as a direction, not an exact change.

Poll or report	Date	Sample	Result	Source
AI governance or formal AI policy	Jan 20, 2026	328 responses	20% have one, 22% developing, 56% none, 2% unsure	MGMA Stat
Formal AI governance structure	Fall and winter 2024	Not stated	73% had none	MGMA and Humana, as cited by MGMA Stat
Homegrown AI app or workflow tool	Sep 22, 2026	228 responses	34% built one, 8% tried and did not launch	MGMA Stat
AI used in patient visits	Aug 5, 2025	244 responses	71% use it, 29% do not	MGMA Stat

Use is ahead of policy

MGMA's August 5, 2025 poll of 244 practice leaders found that 71% used AI in patient visits and 29% did not. Among the users, 47% used it in a quarter of visits or fewer. Another 10% used it in 26% to 50% of visits, 7% in 51% to 75%, and 7% in 76% or more.

A September 30, 2025 MGMA Stat poll found that 68% of medical groups had added or expanded AI tools during 2025. At least 68% of groups were adding tools while 56% had no written rules for them.

The same August 2025 poll asked whether AI reduced staff workload. 44% said it had not, 39% said it had, and 17% were unsure. MGMA noted that groups using AI in more than 25% of visits were more likely to report less workload. A policy that names an owner and a review date gives a practice a way to check that claim against its own numbers.

One in three groups has built its own tool

On September 22, 2026, MGMA Stat asked leaders whether their practice had built its own app or workflow tool with AI. Of 228 applicable responses, 34% said yes and 8% tried but did not launch. 54% said no and 10% were unsure.

Added together, 42% of groups have tried, which MGMA describes as more than 4 in 10 practices. Respondents named tools for scheduling, prior authorization, denial work, dashboards, inventory, call handling, recalls, staff tracking, charting support and coding guidance.

Homegrown tools raise a question a vendor contract would normally answer. MGMA's article points readers to HHS guidance on HIPAA, cloud computing and business associates, to the ONC health IT certification process, and to FDA guidance on clinical decision support from January 2026. Only 20% of groups had a formal policy in January to say who checks those items before a tool goes live.

Where front offices aim automation

MGMA's February 10, 2026 poll of 177 practice leaders asked where the practice focused AI or automation in the front office. Scheduling led at 31%, followed by calls at 27%, registration and eligibility at 23%, and prior authorization at 16%. Other areas made up 5%.

Patient-facing adoption runs slower than the interest. In an April 2025 MGMA poll, only 19% of groups used a chatbot or virtual assistant for patient communication. In a July 2025 poll, 71% of practices had fewer than a quarter of patients using digital tools to self-schedule, and 3% had more than 75%.

These are the areas where a policy has the most to say. Scheduling, calls and eligibility touch patient data every day, so they are the first workflows worth writing rules for.

What a practice manager can take from the numbers

A written policy is a short document. MGMA's January article points to the NIST generative AI profile and to guidance from the Federation of State Medical Boards as starting frameworks, and it stresses human accountability for AI output.

Based on the polls above, a practice that wants to match the 20% with a policy in place needs a list of every AI tool in use, a named owner for each, and a review date. Annual review is the most common cycle among groups that already have a policy.

If your group sits in the 34% that built a tool, add that tool to the list first. If it sits in the 71% that use AI in patient visits, record the share of visits it touches, since MGMA's own data links wider use with more reported workload relief.

MGMA's polls each draw a few hundred responses from its own audience, so they describe MGMA Stat respondents and not every practice in the country. The largest sample above is 328 leaders.

Sources

Talk to us

Fifteen minutes about your practice and the systems you run.

Book a demo