While Washington is weighing how to regulate health care AI, the companies bringing these new technologies to market face a tougher test: proving they work, deliver clinical value and cost savings, and can close a trust gap with patients, clinicians and health systems.

Two recent developments underscore that tension. The Department of Health and Human Services (HHS) finalized a rule that eliminates certain AI transparency requirements, while outgoing Senate HELP Committee Chair Bill Cassidy unveiled a draft framework for how health care AI is paid for and regulated.

The midterm election results add another new variable. With Democrats taking control of Congress in January, health care AI is likely to face new questions and more oversight from Capitol Hill even as the administration weighs its options. But divided government rarely produces new laws quickly, so federal rules may stay unsettled for some time. Companies must earn trust now, with fewer federal standards to lean on and more scrutiny of how they do it.

Transparency Is Now a Market Differentiator

HHS’s new rule removes required disclosures on how certain AI decision-support tools in certified health IT systems, such as electronic health records, are developed, evaluated and used. What was a compliance requirement is now a competitive opening.

Hospitals, physicians, and several medical societies pushed to keep the disclosures, warning that without them, the work of vetting AI tools would fall on providers. Regulators countered that the disclosures hadn’t measurably improved patient care.

Both may be right. Health systems still need to know what they’re buying and the prior disclosures provided little clarity. The same is true across the market: payers want proof of value; employers want evidence before offering AI tools to their workforce; and consumers want to know how AI uses their data. Companies that provide each audience evidence it can act on will stand out.

What Health Care AI Will Have to Prove Next

Cassidy’s draft, the PATIENT AI Act, may never become law, but it raises three questions every health care AI company will face, whoever writes the rules.

1. Who is accountable when AI gets it wrong? The draft would clarify which agency oversees each type of AI tool, reaffirm existing federal limits on some state lawsuits against software-based medical devices, but leave the broader question of liability to further study. For consumer-facing tools that aren’t regulated as devices, the answer is even less settled. Companies that define their own accountability now will be better positioned when Washington does.

2. What evidence earns trust? As federal disclosure rules recede, accreditors and states are stepping in with standards of their own. The Joint Commission, the nation’s leading hospital accreditor, now offers voluntary certification for health systems that govern and monitor the AI they deploy. States are writing rules for consumer-facing AI, from chatbot disclosures to limits on AI in mental health care. Each sets expectations that companies will have to meet.

3. How will reimbursement for AI evolve? The draft would set up a process for Medicare to decide whether and how to pay for AI services, including a pilot to test paying for AI that performs certain clinical tasks on its own. The business question is clear: what evidence of benefit, safety and value will justify reimbursement?

What Companies Should Do Now

Whether you’re a technology platform, consumer health company, clinical workflow vendor or diagnostics developer, three priorities stand out:

  • Own the company story. Product claims aren’t enough. Be precise about what your technology does, where it fits in care, how it approaches safety and why it should be trusted. Communicate that consistently to regulators, customers, investors and the public before others define it for you.
  • Know your stakeholders. Translate your evidence. Map who shapes adoption, from clinicians and health systems to payers, policymakers and patients, and what each needs to see. Then communicate your evidence on real-world performance, safety and impact in terms each audience can act on.
  • Engage where the rules are being made. While Congress debates, states are acting: by mid-year, 20 states had enacted AI laws affecting health care, including seven targeting consumer-facing chatbots. Accreditors, health systems and payers are setting standards too. Companies that bring their evidence to these debates, including Cassidy’s draft (open for comment until December 1), can help shape the rules instead of reacting to them.

The Bottom Line

The rules for health care AI are still being written, and a divided Washington may pull them in different directions. The market’s questions are more settled: Does it work? Where does it fit? Can it be trusted? Companies with credible answers will be ready for whatever comes next.

At Reservoir Health, we help companies bringing AI into health care understand the policy forces shaping their markets, identify stakeholder concerns and develop positioning and communications that explain their value and earn trust.

‍