:max_bytes(150000):strip_icc():format(jpeg)/GettyImages-2259532243-5f2bc9bd08d6492da1499eff6ddfd95f.jpg)
KEY TAKEAWAYS
- The Centers for Medicare and Medicaid has launched a trial program in six states that uses AI to inform claims decision-making.
- Claims that need prior authorization will go through AI first. Claims for emergency inpatient-only services and emergency services will not be processed by AI.
- CMS says the updated technology will decrease waste and speed the prior-authorization process. Some experts and lawmakers are worried it will deny claims for services that should be covered by Medicare.
Your next Medicare claim could be approved or rejected by AI.
Last month, the Centers for Medicare and Medicaid Services (CMS) introduced a new system that will use AI and machine learning to approve or deny some Medicare claims. The temporary cost-cutting program, called the Wasteful and Inappropriate Service Reduction, or the WISeR Model, will run through the end of 2031 and operate in six states: Texas, New Jersey, Oklahoma, Ohio, Washington and Arizona.
Typically, a hospital or doctor will submit requests to Medicare for prior authorization. Non-emergency health care generally requires prior authorization, which allows the provider and patient to know whether the service will be covered by Medicare. It is generally required for services like a planned surgery, an MRI scan, or services and medicines with side effects or a higher potential to harm the patient.
CMS said pre-authorization claims submitted through the WISeR Model will be denied if the medical service provides little to no clinical benefit or has a “higher risk of waste, fraud, and abuse.” For example, skin and tissue substitutes, electrical nerve stimulator implants, and knee arthroscopy for knee osteoarthritis are usually rejected as they are more likely to waste Medicare finances and can cause physical harm to the patient.
“AI could help Medicare process prior authorizations and claims more quickly and consistently, reducing delays and administrative burdens for beneficiaries and aiding in the fight against fraud and waste,” Shannon Benton, the executive director of The Senior Citizens League, said in an email. “However, risks are significant, as automated systems might wrongly deny or delay necessary care, especially in complex cases.”
Why This Matters
If using AI reduces unnecessary care, it could save Medicare millions of dollars and ease pressure on the federal budget, with positive implications for federal deficits, taxes, and interest rates. But if the system increases denials or disputes, it could shift costs onto seniors, who are generally on fixed incomes in retirement.
Who Is Impacted and Who Isn’t
Most Medicare prior authorization claims will go through the WISeR program if the beneficiary lives in a state that’s using it.
This program does not change what is and isn’t covered for Medicare beneficiaries, and claims for inpatient-only services and emergency services will not go through the WISeR Model, the CMS said.
The new model also applies only to beneficiaries with Original Medicare. The roughly 54% of beneficiaries with coverage through a Medicare Advantage plan will not be affected. Some Medicare Advantage insurers have already started using AI to assist with claims decisions, but lawmakers argue that this has contributed to the mismanagement of the process.
The Risks of the WISeR Model
CMS says implementing the WISeR model will help expedite claims decisions and save the agency money by preventing unnecessary or harmful claims. It adds that any claims that are denied by AI will be reviewed by a qualified clinician before a final decision is issued.
“Any use of AI would be subject to strict oversight to ensure transparency, accountability, and alignment with Medicare rules and patient protection,” a CMS spokesperson said in an email. “CMS remains committed to ensuring that automated tools support, not replace, clinically sound decision-making.”
The spokesperson added that CMS will monitor the timeliness and accuracy of prior-authorization review decisions.
Objections to the Use of AI by Medical Professionals
Lawmakers, experts and beneficiaries, however, are wary of the new model. Medicare is one of the last types of health insurance in the country to bring AI into the claims decision process. Many private insurers are already using AI.
Last year, about three in five physicians surveyed by the American Medical Association were concerned that AI will “increasingly override good medical judgment” and hurt patients by increasing claim denials.
After plans to introduce the WISeR model were announced last year, a group of House lawmakers wrote to CMS requesting that the program be canceled, as it will likely limit access to necessary health care and create incentives to “put profit over patients,” the letter said.
Medicare’s prior authorization decisions are typically outsourced to a company that uses technology, such as AI, to recommend whether a claim should be approved or denied. And the agency confirms that companies that use the WISeR model will “receive a percentage of the expenditures associated with averted wasteful, inappropriate care as a result of their reviews.”
In a recent poll of Medicare beneficiaries by The Senior Citizens League, 53% opposed “combining AI with human review for timely Medicare payments,” Benton told Investopedia.
“Beneficiaries need transparency, clear explanations for denials, and strong appeal rights when AI is involved,” Benton said in an email. “The [WISeR model] could create additional barriers to care, particularly for seniors with serious or chronic conditions. Any AI use must include safeguards and human oversight to ensure access to needed treatment.”
No matter how a claim is denied, a health care provider or beneficiary can still appeal if they believe it was in error. After a claim is processed, beneficiaries will receive a Medicare Summary Notice by mail or electronically. The notice will inform them whether their claim was denied or approved and how to appeal the decision.

:max_bytes(150000):strip_icc()/GettyImages-2259532243-5f2bc9bd08d6492da1499eff6ddfd95f.jpg)