Here is the paraphrased version of the provided HTML content:
Terry Gerton I’m glad to have you here. The Bipartisan Policy Center has released a new report we’ll be discussing: Understanding the Regulation of Health AI Tools. When most people hear “health AI tools,” they probably picture a fitness tracker or wearable device. What exactly does this report focus on?
Maya Sandalow Great question. My organization, the Bipartisan Policy Center, has worked extensively to clarify this very issue. AI is being used across healthcare in numerous ways. To keep it simple, I break it into two main categories. The first is administrative AI, which handles operational tasks. As a patient, you might use it for appointment scheduling or understanding your benefits. One of the fastest-growing uses is AI-powered ambient scribes, which help doctors record patient visits. Health insurers also use it for claims processing, particularly for prior authorization. That covers the operational side. The second category is clinical AI. This includes traditional medical devices—tools designed to diagnose, prevent, or treat illness. It also includes consumer wearables like smart rings and Apple watches. The range is vast, and even within medical devices, applications vary widely. So far, the most common use is in medical imaging, but we can explore all these areas in more detail.
Terry Gerton That’s a helpful overview. But why is regulation even an issue? What makes it so challenging to regulate these different types of health AI?
Maya Sandalow The regulatory challenges vary significantly depending on the type of AI. The Food and Drug Administration (FDA), part of the Department of Health and Human Services (HHS), oversees medical devices to ensure they’re safe and effective before reaching patients. The same standard applies to AI used in clinical settings—we need to be confident it’s safe and reliable before it’s used on someone. Beyond that, there may be a need for oversight when AI is used to determine whether a medical procedure will be covered by insurance. So, the policy questions differ based on the AI’s specific function.
Terry Gerton One key finding from the report is that many of these tools aren’t classified as medical devices, making it unclear who’s responsible for regulating them—or whether multiple agencies share that responsibility. What does the current regulatory landscape look like?
Maya Sandalow It’s definitely complex. The same AI tool might be regulated by different bodies depending on where it’s used and what data it’s trained on. We’ve mapped out the main federal agencies involved. The Centers for Medicare & Medicaid Services (CMS) handles coverage and reimbursement decisions for AI in clinical care—essentially, how much providers or patients get paid for using these tools. The Office for Civil Rights at HHS enforces HIPAA, which is crucial for patient privacy. Any AI used in traditional healthcare settings involving patient data falls under their oversight. However, this may not apply to consumer wearables like smartwatches or rings, which exist outside the traditional healthcare system. The Federal Trade Commission can step in if there’s misleading marketing about a tool’s intended use. Then there’s the Office of the National Coordinator for Health IT—quite a mouthful—which certifies electronic health records. When AI is built into those systems, this office has jurisdiction. So developers and providers must navigate a complicated web of federal regulations, and that’s just at the national level. States are also stepping in, meaning a tool might face strict rules in one state and none in another.
Terry Gerton Maya Sandalow is associate director of the Bipartisan Policy Center’s health program. Maya, what you’ve described goes far beyond what most people imagine. It seems like AI is advancing in healthcare faster than the rules can keep up. Everyone’s adopting AI in some form, but the regulations aren’t fully in place yet. What risks or gaps does this create?
Maya Sandalow That’s a critical question. One major trend to note is that right now, most AI in healthcare is used for administrative tasks—like doctors transcribing notes or insurers reviewing claims to determine coverage and reimbursement. The clinical potential of AI is enormous: early disease prevention, more accurate diagnoses, and better treatment of complex conditions. But the regulatory environment is so complicated, and the payment structures are unclear, that it’s creating a kind of chilling effect. There’s a big policy challenge ahead: how do we build regulatory and payment frameworks that ensure safe, effective, high-value AI tools actually reach the patients who need them most?
Terry Gerton Who should take the lead in creating that policy framework?
Maya Sandalow Right now, many states are exploring this—after all, states are often called laboratories of democracy, so we may see best practices emerge. But these technologies operate nationally, with developers trying to deploy tools across the country. Many believe a federal framework is needed. HHS, which oversees much of healthcare delivery, is actively working on this right now.
Terry Gerton Meanwhile, the profit opportunities for companies often lie in those regulatory gray areas. How are industry incentives influencing the regulatory process?
Maya Sandalow This is incredibly important. The goal should be to create a
Maya Sandalow The idea is to build a set of rules for developers and tech companies that encourages them to lean into meeting regulations instead of trying to sidestep them. As things stand, the process of getting through the regulatory system can be incredibly complex. At the same time, the benefits of compliance aren’t always clear. This creates a strong motivation for companies to claim that their creations—say, a software innovation—aren’t medical devices, and therefore shouldn’t fall under FDA oversight. This is a hotly debated issue. For instance, users of AI-powered chatbots or tools might not be aware that these products aren’t regulated for diagnostic or treatment purposes. We need a system where developers are actively encouraged to comply with regulations so that patients are assured of using safe and effective tools.
Terry Gerton The emphasis on “safe and effective” is key, because another role of the regulatory system is to determine who is held responsible if something goes wrong.
Maya Sandalow Exactly. That’s a major concern right now. Often, healthcare providers are the ones held accountable when problems arise from using AI in patient care. States are actively debating this issue. An important recent policy move to note is from the Office of the National Coordinator for Health IT (ONC), which a few years back introduced a “nutrition label” or model card policy. This required developers to provide transparent information about their AI systems—similar to a nutrition label—so that healthcare providers could understand the risks and share responsibility accordingly. In early January, the new administration floated a proposal to repeal this rule. While opinions are divided, many providers are worried that rolling it back would leave them without the necessary information to make informed decisions about AI use and could expose them to greater liability.
Terry Gerton Given this state of regulatory uncertainty, what should individuals know? How can they protect their privacy and data while still making the most of available AI tools?
Maya Sandalow That’s a great question. The simplest advice is: don’t assume that an AI-driven solution you encounter is regulated in the same thorough way as prescription medications. That doesn’t mean these tools aren’t useful or powerful—just understand that they aren’t substitutes for professional medical advice. It’s also wise to discuss with your own doctor the AI tools you’re using so you can jointly decide how best to integrate them into your care.
Terry Gerton What’s next for federal regulation in this area?
Maya Sandalow This is definitely something to watch. There’s broad bipartisan support for a unified federal regulatory framework, mainly because relying on differing state-by-state rules has become unwieldy. The challenge lies in the specifics—where consensus is harder to achieve. Still, I expect this conversation to continue in the coming years: the push for a national framework, along with ensuring that executive agencies have both the authority and the resources needed to play a larger role in overseeing AI in healthcare.
Copyright © 2026 Federal News Network. All rights reserved. This website is not intended for users located within the European Economic Area.
