Posted on by The ACG News Team
Stephen T. Amann, MD, FACG

Legislative & Public Policy Update from Chair Stephen T. Amann, MD, FACG

We’ve faced this kind of chaos before.

At the beginning of the 20th century, American industry was a patchwork: bricks and screws of different sizes, broadcasters inadvertently sharing the same radio frequencies, and even traffic lights without the now-ubiquitous ‘red-yellow-green’ color scheme.

In the 1920s, Secretary of Commerce Herbert Hoover—better known as the president during the Great Depression—brought together scientists, policymakers, and industry leaders to develop many national standards we now take for granted. Hoover created a more predictable environment that both encouraged innovation and growth and saved lives.

Fast forward to today, the rapid deployment of artificial intelligence presents a similar challenge. In healthcare alone, clinicians, payers, researchers, hospitals, and technology companies are increasingly relying on AI tools for both administrative and clinical purposes. And with the recent news that AI models have broken out of restricted testing environments and hack other companies’ data, the risk of autonomous cyber threats is an emerging concern. Yet, there remains no comprehensive framework governing how these technologies are developed, evaluated, or deployed.

Unlike Hoover’s era, the question today isn’t if someone will step up to create these standards, but who. In recent years, the Executive Branch, Congress, and the states have all moved to claim this role. The result is a growing debate that defies the typical partisan lines over whether the country should adopt a single national standard or continue allowing states to develop their own approaches. The outcome of that debate—and how to best balance innovation, safety, accountability, and patient protection—will affect everyone involved in U.S. healthcare.

Federal Government: The Administration & Congress

The Trump Administration

Since returning to office, President Trump has made AI leadership a central policy priority. Like with other industries, the administration has generally favored a light-touch regulatory approach, designed to accelerate innovation and maintain U.S. competitiveness.

The White House has also advanced their view that a national framework must take precedence over a growing patchwork of state AI laws; in a December executive order, the administration expressed its goal to work with Congress in creating a “minimally burdensome” national, uniform framework.

In June, President Trump issued a separate executive order for a voluntary pre-deployment review of technology to assess the cyber capabilities of frontier AI models (e.g., the most advanced general-purpose models).

However, there has already been bipartisan pushback from the states. For example, Utah Governor Spencer Cox (R-UT) criticized the Trump Administration after it formally opposed his state’s efforts to require AI companies publish safety and child protection plans. Following this, more than 50 Republican state legislators from 22 states said they were “deeply concerned” by the administrations’ desire to block state AI policymaking. Various Democratic governors have also introduced AI-related executive orders in recent months, or worked with their state legislature to pass laws on AI, deciding to move forward with enacting state policies and not waiting for the Trump Administration to implement uniform standards.

Congress

Unsurprisingly, Congress has struggled to reach consensus on any comprehensive AI legislation. While some lawmakers agree that some level of federal oversight is necessary, disagreement remains over the scope of regulation and the extent to which federal law should preempt state action.

Last year, Congressional Republicans considered a 10-year moratorium on state AI laws as part of their One Big Beautiful Bill. However, after significant, bipartisan pushback from governors and state attorneys general, Republicans on the Hill backed away from these plans.

In July, Representatives Jay Obernolte (R-CA) and Lori Trahan (D-MA), introduced the bipartisan Frontier Risk Oversight, National Transparency, Independent Evaluation, and Reporting (FRONTIER) Act, H.R. 9925. The bill implements a unified national framework that regulates the development and deployment of cutting-edge AI models based on their level of risk. Although the bill would prevent states and local governments from creating new requirements for AI developers in specific frontier AI safety areas, it does preserve the authority of states to regulate AI in some areas (e.g., protecting minors, use of the technology, and procurement).

What does the absence of a national standard mean for GI practices?

If the White House and Congress succeed in establishing a national framework, GI practices could face a more uniform set of AI requirements across states. But ACG members know all too well that reaching a speedy consensus in Washington is unlikely.

Without legislation, the federal government could effectively create a national AI standard by leveraging its control over Medicare, Medicaid, ERISA, and the VA, which cover millions of Americans on those insurance plans. The administration could also advance their AI goals by seeking voluntary industry commitments or using existing regulatory frameworks (e.g., the FDA clears AI tools, like CADe/CADx polyp detection, through its Software as a Medical Device classification).

States

With states increasingly take the lead in regulating AI, the result is a rapidly expanding and diverse collection of laws and executive orders that vary significantly in scope, focus, and regulatory approach. For healthcare organizations operating across multiple states, maintaining compliance may become increasingly complex—perhaps confirming the fears about a patchwork of AI laws without common standards.

The National Conference of State Legislatures (NCSL) estimates that in 2025, legislators in every state introduced AI-related legislation, with 38 states ultimately passing 145 laws.  By all indications, this trend accelerated in 2026.

In healthcare, state laws tend to focus on setting standards for human roles in decision-making and on transparency. Trends include oversight of insurers’ use of AI (particularly for prior authorization and utilization reviews), requiring clinicians to inform patients when AI is used to assist in diagnosis or treatment, establishing regulatory sandboxes to test healthcare AI innovations before wider deployment, and restrictions on the use of AI in therapy or behavioral health.

States argue they are the ‘laboratories of democracy,’ but AI is a perfect example, with states both encouraging and restricting use of the technology. Perhaps the best-known initiative to emerge from a regulatory sandbox is Utah’s Doctronic pilot project, which has raised alarms for permitting AI systems to autonomously review and approve routine prescription refill requests for a limited set of patients with chronic conditions. States may also consider adopting licensure pathways; for example, the conservative Cicero Institute has developed model legislation that would classify AI models as licensed healthcare entities rather than as medical devices. Meanwhile, seven other states have passed laws in 2026 that limit or prohibit insurers from using AI in prior authorizations.

Interested in what’s happening in your state? The NCSL maintains a database tracking AI legislation across state legislatures, including AI funding, government use, private-sector use, AI in healthcare, responsible use and discrimination, studies, and more.

What does this mean for GI practices?

Health insurance and the practice of medicine is still largely regulated at the state level. States can pass legislation and implement policy changes much faster than the federal government, which for AI, could be a double-edged sword. On one hand, states are better positioned to match the speed of a rapidly evolving industry; on the other, practices operating in multiple states may need to navigate a confusing, inconsistent maze of rules governing clinical decision support, patient disclosures, prior authorization, and future autonomous AI applications.

What is ACG doing for you?

As we’ve covered, policymakers at both the federal and state levels are making decisions that will shape how AI is developed, validated, deployed, and monitored in clinical practice for years to come.

For ACG members and GI practices, emerging laws could affect clinical decision support tools, documentation and coding systems, prior authorization processes, patient communications, and future autonomous AI applications. Depending on where you practice, the regulatory requirements governing these technologies may differ considerably.

Above all, with federal and state policymakers continuing to debate the appropriate balance between innovation and oversight—and who gets to set those rules—physicians must remain engaged and informed. ACG is tracking this fight through our Board of Governors, committees, and relationships with policymakers at both levels, so that when the Herbert Hoover of AI emerges, we’ll be at the table to represent the best interests of GI patients and practices.

ACG would like to thank Sheila Madhani (Madhani Healthcare Consulting, LLC) for her contributions in drafting this article.