
Share
At HLTH in Las Vegas, health tech leaders are debating how Medicare should regulate artificial intelligence in care delivery, a question with real consequences for patients who may never know an algorithm touched their treatment.
Imagine your doctor's office humming with software that reads your scans, predicts your risk of readmission, and drafts your discharge instructions. Now imagine nobody has agreed on the rules for how that software gets paid for, checked for errors, or held accountable when it gets something wrong. That's roughly where American health care stands right now, and it's the backdrop for this week's gathering at HLTH, one of the industry's biggest annual conferences, held this year in Las Vegas.
Thousands of hospital executives, insurers, startup founders, and policymakers descended on the city to hash out where health technology is headed. Among the central topics: how the Centers for Medicare and Medicaid Services, the federal agency that controls payment for roughly 150 million Americans' health coverage, should approach artificial intelligence. CMS sets the financial incentives that shape what kind of care actually gets delivered in this country. If the agency decides it will pay for an AI-assisted diagnosis or an algorithm-driven care plan, that technology spreads fast. If it doesn't, even the most promising tool can stall in pilot purgatory.
That's why so much attention is focused on what's being called a CMS roadmap for AI, a framework meant to clarify how the agency will evaluate, cover, and oversee artificial intelligence tools used in the care of Medicare and Medicaid beneficiaries. For an industry that has spent years building AI products without a clear sense of how they'll be reimbursed, a roadmap like this is less a formality than a starting gun. Startups and hospital systems alike have been investing ahead of regulatory clarity, a risky bet in any sector but especially so in health care, where the margin for error touches people's bodies and lives.
Think of a regulatory roadmap the way you'd think of a building code. It doesn't tell you exactly what kind of house to build, but it tells you what standards your house has to meet before anyone will insure it or let you move in. Health AI has been operating for years without that kind of code. Developers have built diagnostic tools, administrative automation, and clinical decision-support systems that vary wildly in quality, transparency, and evidence of safety. Some have gone through rigorous clinical validation. Others have been deployed with far less scrutiny, leaning on hospital systems eager to cut costs or improve efficiency.
A clearer federal framework could change that calculus. It could set expectations for how much clinical evidence a tool needs before it gets reimbursed, how its performance should be monitored once it's in use, and how patients or clinicians can flag problems. That matters enormously for equity. AI models trained on unrepresentative data have a well-documented history of underperforming for women, for Black and Hispanic patients, and for people with rarer conditions. Without oversight baked into the payment system itself, there's limited incentive for companies to fix those blind spots once a product is already generating revenue.

There's also a flip side worth naming honestly. Overly rigid rules could slow down tools that are already helping clinicians catch things they'd otherwise miss, like early signs of sepsis or subtle patterns in imaging that predict disease progression. The challenge CMS faces is calibrating oversight so it filters out unsafe or unproven tools without choking off innovation that's already improving outcomes in limited settings. That's a genuinely hard needle to thread, and it's one reason conversations at HLTH this year carried more urgency than in past years, when AI discussions tended toward speculative enthusiasm rather than concrete policy questions.
Electronic health records, the backbone of how American medicine stores and shares patient information, were another flashpoint at the conference. A STAT-hosted session in Las Vegas dug into what's being called a dose of reality on EHRs and AI, reflecting a broader mood shift in the industry. After years of hype about AI transforming clinical documentation and reducing physician burnout, there's growing acknowledgment that integration into existing records systems remains clunky, expensive, and uneven across hospital systems. The gap between a slick product demo and a functioning tool inside a chaotic emergency department workflow is often wider than vendors admit.
None of this unfolds in a vacuum. The same week, a federal judge issued a final ruling blocking a U.S. plan to limit the length of visas for international graduate students and postdoctoral researchers, a policy that would have squeezed the pipeline of scientific talent many health tech and biotech companies depend on. Elsewhere, the United States lost its measles elimination status, a stark reminder that even as the industry debates futuristic algorithms, basic public health infrastructure is fraying in ways that predate any AI conversation. These threads aren't unrelated. A health system distracted by eroding public health fundamentals and workforce uncertainty has less bandwidth to carefully vet the AI tools flooding into its workflows.
The stakes here aren't abstract. Every policy decision CMS makes about AI coverage ripples outward into exam rooms, nursing stations, and the kitchen tables where families try to make sense of a diagnosis. A well-designed roadmap could mean faster, more accurate detection of disease for patients who currently wait weeks for specialist review. A poorly designed one could entrench bias, waste public money on unproven tools, or leave patients without recourse when an algorithm gets something wrong.
What's striking about HLTH this year is how much the conversation has matured. Fewer people are asking whether AI belongs in health care. Most seem to accept it's already here. The harder, more consequential question is who gets to decide what counts as safe, effective, and fair, and whether the agencies responsible for that decision move quickly enough to keep pace with an industry that shows no sign of slowing down. For patients, that question will determine not just what kind of care they receive, but whether they can trust it.
Tags
Original Sources
What to know about the new AI roadmap from CMS, and news from HLTH
↗ https://www.statnews.com/2026/11/17/cms-ai-roadmap-and-news-from-hlth-health-tech
About the author
Amara's entry point into AI was an epidemiology role at a London research hospital, where she spent five years studying how digital health tools reached — or conspicuously failed to reach — underserved communities. Watching early algorithmic systems in healthcare quietly entrench existing inequalities, she redirected her career toward the systemic consequences of AI at scale. She covers AI through an unflinching lens: who benefits, who bears the cost, and what evidence actually says versus what the press release claims. Her writing is calm and precise, but she doesn't mistake balance for neutrality.
More from The Steward →This Week's Edition
18 November 2026
29 articles
Related Articles

Merck Taps Isomorphic Labs for AI-Designed Oncology Drug in Latest Pharma-AI Tie-Up
Health & Science · 5 min

FDA Approves First Drug to Treat Brain Symptoms of Type 3 Gaucher Disease
Health & Science · 6 min

Stanford Pairs AI With Brain Science to Tackle Depression, Memory, and How the Mind Develops
Health & Science · 6 min
Related Articles

Merck Taps Isomorphic Labs for AI-Designed Oncology Drug in Latest Pharma-AI Tie-Up
Health & Science · 5 min

FDA Approves First Drug to Treat Brain Symptoms of Type 3 Gaucher Disease
Health & Science · 6 min

Stanford Pairs AI With Brain Science to Tackle Depression, Memory, and How the Mind Develops
Health & Science · 6 min
More Stories
© 2026 Cedar & Bloom. All rights reserved.