Estimated reading time: 6 minutes · Last updated:
The Centers for Medicare and Medicaid Services (CMS) has published a clinical AI roadmap setting out how the agency wants artificial intelligence to enter American medicine. The plan rests on four priorities: earning public trust, improving the systems that move patient data, building risk-based regulatory routes with the Food and Drug Administration, and paying for technology by the results it produces for patients rather than the volume of services it generates. CMS released the framework on a Friday in November 2026, and Nextgov/FCW covered it the following Monday. The roadmap names the risks it wants to manage, including patient safety, data privacy, cybersecurity and algorithmic bias. It sets no deadlines, so the regulatory and payment follow-through is what will decide whether it matters.
Key takeaways
- Four priorities: The CMS clinical AI roadmap rests on public trust, stronger health data infrastructure, clearer regulatory pathways and outcomes-based payment models.
- FDA coordination: CMS plans to work with the Food and Drug Administration on risk-based rules, including a possible competency-based evaluation for some AI-enabled medical devices and quicker market routes for lower-risk tools.
- Payment reform: CMS cited its ACCESS model, plus the existing LEAD and AHEAD models, as routes for rewarding technology-enabled care that improves outcomes and lowers costs.
- Industry backing: More than 60 companies, including Oracle, Apple, Amazon, Google, Anthropic and OpenAI, pledged support for a patient-centred digital health ecosystem at a White House event in August 2025.
Table of contents
- Key takeaways
- Why Trust and Data Come Before Scale in the CMS Plan
- How CMS and the FDA Plan to Regulate AI by Risk Level
- Can Outcomes-Based Payment Change What AI Gets Built
- CMS Is Measuring Its Own AI Use by Results
- What could help or hold back the roadmap
- What to be careful about
- Frequently asked questions
Why Trust and Data Come Before Scale in the CMS Plan
The first priority is public trust, and CMS treats it as a question of pace. The agency wants clinical AI introduced in stages. Each stage would be backed by published performance results, patient consent and monitoring that continues after a tool goes live. Staged adoption means a technology earns wider use only after it has shown how it performs in narrower settings.
The second priority concerns data. A clinical AI model is only as useful as the records it can read, and CMS wants three things improved. Patients should get broader access to their own medical records. Information should move more easily from one provider to another. And the channels carrying that information should be secure.
Taken together, these two priorities put the groundwork ahead of the product. An algorithm that cannot reach a complete patient history has little to work with, however capable it is. The same holds for a tool that patients will not trust with their records. CMS is betting that adoption follows confidence and access, not the other way round.
How CMS and the FDA Plan to Regulate AI by Risk Level
The third priority is regulatory. CMS says developers need clearer national standards and a risk-based approach if they are to bring validated tools to market. A risk-based approach scales oversight to the harm a product could cause, so a consumer wellness app would not face the same scrutiny as software involved in clinical care.
CMS plans to build that system jointly with the Food and Drug Administration. For some medical devices that rely on AI, the roadmap floats a method of evaluation grounded in demonstrated competency, presenting it as an option under consideration rather than a settled approach. It also sets out quicker paths to market for consumer wellness products and clinical decision support applications that carry lower risk.
The two agencies want stronger post-market surveillance as well. That means collecting real-world performance data once a product is in use, so regulators can check whether a technology that keeps evolving remains safe and effective. For AI, which can behave differently after an update, monitoring after approval may matter as much as the gate before it.
Can Outcomes-Based Payment Change What AI Gets Built
The fourth priority targets incentives. CMS says it will explore reimbursement structures that reward technology for better patient outcomes and lower healthcare costs. It explicitly sets these against payment for more service volume or more administrative activity. Under that older logic, a tool that generates billable work can earn more than one that keeps patients healthy.
CMS cited its ACCESS model as a working example. Under ACCESS, care for chronic conditions that is delivered using technology becomes eligible for payment opportunities tied to outcomes. The agency also named two existing payment structures, the LEAD and AHEAD models, as potential pathways for supporting care delivered with technology.
The wording is exploratory, and that matters. Saying the agency intends to explore new structures, and calling LEAD and AHEAD potential pathways, is different from changing what those models pay for. Until payment rules move, developers have a stated direction from CMS rather than a revenue model they can build a business on.
CMS Is Measuring Its Own AI Use by Results
The roadmap matches a shift inside the agency. CMS is moving its own AI strategy toward measuring operational outcomes and away from relying mainly on adoption statistics. Patrick Newbold, the CMS chief information officer, explained in September that the agency judges AI as a means of reaching specific organisational goals, not as something worth adopting for its own sake.
Its Tech Pulse initiative puts that into practice. Tech Pulse tracks technology-related objectives and key results, a goal-setting method that pairs each aim with measurable markers of progress, against agency priorities.
CMS has also launched QualTech, an industry engagement effort seeking commercially available technology to address healthcare quality problems. Working through its Center for Clinical Standards and Quality, the agency settled on four priority areas: AI-assisted reviews of patient safety, support for annual wellness visits, measuring quality by digital means, and a nationwide dashboard tracking hospital quality.
Industry interest was already on record. In August 2025, during a health technology event hosted at the White House, upwards of 60 companies committed to supporting a digital health ecosystem centred on patients, resting on interoperable data and care delivered with the help of AI. Oracle, Apple, Amazon, Google, Anthropic and OpenAI were among the signatories.
| Priority | What CMS proposes | Linked programmes or partners |
|---|---|---|
| Public trust | Staged adoption, performance reporting, patient consent, continuous monitoring | None named |
| Data infrastructure | Patient record access, provider-to-provider exchange, secure data sharing | White House pledge companies |
| Regulatory pathways | National standards, risk-based rules, post-market surveillance | Food and Drug Administration |
| Outcomes-based payment | Reward better outcomes and lower costs, not volume | ACCESS, LEAD, AHEAD models |
What could help or hold back the roadmap
The case for
- Coordination with the FDA could give developers of lower-risk wellness and decision support tools a faster route to market.
- The ACCESS model already offers outcome-aligned payment for technology-enabled chronic care, so the payment idea has a live test bed.
- More than 60 companies, including Oracle, Apple, Amazon and Google, have publicly pledged support for interoperable health data.
The case against
- The payment reforms are framed as something CMS will explore, and the LEAD and AHEAD models are only potential pathways.
- The competency-based device evaluation is described as a possibility, with no defined method yet.
- The roadmap names no timetable for any of its four priorities.
What to be careful about
- Faster market routes for lower-risk tools could let flawed clinical decision support software reach clinicians before post-market surveillance catches problems.
- Wider data sharing between providers expands the privacy and cybersecurity exposure that CMS itself lists as a risk.
- Algorithmic bias, which the roadmap names, could persist if performance reporting does not break results down across patient groups.
The bottom line
The CMS clinical AI roadmap is a statement of direction more than a set of rules. Its strongest commitments sit in areas CMS already controls: tying its own AI use to measurable results through Tech Pulse, and seeking quality technology through QualTech. Its most consequential ideas, risk-based approval with the FDA and payment tied to patient outcomes, remain things the agency intends to explore. That gap is the story to follow. If CMS rewrites payment in models such as LEAD and AHEAD, developers will have a reason to build tools that keep patients well rather than tools that add billable work. If it does not, the four priorities will read as good intentions with no money behind them.
What to watch
- Watch for the joint CMS and FDA regulatory approach for clinical AI, including how competency-based evaluation of AI-enabled devices would work; the roadmap gives no date for it.
- Watch whether CMS changes the LEAD or AHEAD models to reward technology-enabled care by outcomes; the roadmap names no timetable.
- Watch for which commercially available technologies CMS selects through QualTech for its four quality priority areas; no selection date has been published.
Frequently asked questions
What is the CMS clinical AI roadmap?
It is a framework from the Centers for Medicare and Medicaid Services for bringing clinical artificial intelligence into healthcare. It sets four priorities: public trust, data infrastructure, regulatory pathways and outcomes-based payment.
Does the CMS roadmap change how Medicare pays for AI tools?
Not yet. CMS says it intends to explore outcome-based reimbursement and names the LEAD and AHEAD models as potential pathways, while its ACCESS model already offers outcome-aligned payment for technology-enabled chronic care.
Which companies support the federal digital health push?
Those commitments came from more than 60 companies at an August 2025 health technology gathering at the White House, a list that featured Oracle, Apple, Amazon, Google, Anthropic and OpenAI.
Related reading