The Trump administration said it had confirmed the enrollments were unauthorized. Now insurers say some of those people should never have lost their health coverage.
On August 31, the Centers for Medicare and Medicaid Services canceled about 315,000 ACA enrollments covering more than 760,000 people. CMS said the enrollments were unauthorized and projected that it would recover about $2.2 billion in premium tax credits. In September, Vice President JD Vance called the broker-created sign-ups “phantom people,” NPR reported.
On October 15, KFF Health News reported, in a story carried by NPR, that insurers had confirmed a small number of reinstatements for people whose coverage should never have been canceled. The total number of wrongful cancellations remains unknown. Mehmet Oz’s CMS announced an enormous enforcement success without giving the public the information needed to measure its mistakes.
Suspicion became a cancellation
The screening combined several warning signs. A broker had submitted the application without a Social Security or immigration document number. Subsidies covered the entire premium. There were no medical claims and no recorded contact with the insurer. Katie Keith’s Health Affairs analysis describes the criteria and the 30-day response window. KFF Health News subsequently reported that insurers had to use at least two contact methods before CMS canceled coverage for nonresponders.
Taken together, those signs can justify investigation. They cannot establish that every person who fails to answer was enrolled without permission. Keith’s analysis explains how a process intended to investigate unauthorized coverage became the basis for mass cancellation. The distinction matters because silence tells an insurer much less than an explicit statement that someone never requested the policy.
CMS itself recognizes the limits of its screening criteria. In its own September rule, the agency acknowledges that having no claims does not, on its face, establish unauthorized enrollment. Healthy people can go a long time without filing one. Leighton Ku’s separate Health Affairs analysis also explains why people who use little care still need insurance against an unexpected illness or accident.
I want an agency canceling health coverage to distinguish an unauthorized policy from an authorized policy whose owner missed the outreach. CMS’s announcement treats its conclusion as settled. The reinstatements reported by insurers show that it was wrong in at least some cases. Nothing in the available evidence establishes how large that group is, and nobody should pretend otherwise in either direction.
The fraud estimate swallows the price increase
The administration also wants credit for a much larger enrollment decline, based on evidence that cannot carry that conclusion.
A June HHS report estimates that improper, phantom, and fraudulent enrollment reached 5.6 million people in 2025, and credits federal integrity measures with removing or blocking 2.9 million. Its estimate of remaining suspect enrollment depends on a model of how much enrollment growth other factors explain relative to a 2019 baseline. It is not a count of millions of individually proven fraud cases. Even its categories differ: a person enrolled by a dishonest broker without consent is a victim of fraud, not its perpetrator.
Meanwhile, KFF’s analysis of 2026 costs found that average monthly premium payments rose from $113 to $178, a 58 percent increase, after enhanced tax credits expired at the end of 2025. Average deductibles rose 37 percent to $3,786, largely because people shifted toward plans with higher deductibles.
KFF Health News reported in August that enrollment had fallen by nearly 3 million and that independent policy analysts disputed the administration’s attempt to attribute the decline to improper enrollment. An aggregate model cannot tell us that nobody left because coverage became unaffordable. Calling the lost customers improper does not make their premium bills disappear.
Oz and Health Secretary Robert F. Kennedy Jr. cannot answer a cost crisis simply by assigning the missing customers to a fraud estimate. The administration needs to show what it actually established about those customers.
Broker enforcement does not validate every cancellation
Stopping unauthorized enrollment protects consumers as well as taxpayers. Taxpayers should not finance policies consumers never requested, and consumers should not have their information used to generate commissions. A cluster of missing identifiers, no claims, and no contact gives investigators a reasonable place to start.
CMS also has taken action against brokers. Its fact sheet lists more than 200 termination notices since January, a moratorium on new broker registrations for 2027, renewed identity checks, and a planned requirement for electronic consumer authorization. Enforcement against dishonest brokers is appropriate.
A program can punish dishonest brokers and still cancel legitimate coverage. Its broker cases do not tell us how many innocent customers it caught.
There are practical alternatives to making consumers repair the damage afterward. In its September statement, Covered California describes requiring verifiable consumer consent before an enrollment partner can access an account or enroll someone. The state marketplace says it has not seen systemic fraud. That self-assessment needs independent scrutiny. The consent requirement nevertheless shows how a marketplace can check permission before a broker acts.
CMS owes the public an error count
KFF Health News reports that another 419,000 sign-ups are under scrutiny. Blue Cross Blue Shield Association official David Merritt said insurers’ reports on nonresponders are due in mid-to-late November, after the midterms. That is a reporting deadline. It does not establish when every cancellation will happen or prove that officials chose it to hide losses from voters.
The accountability problem does not require a theory about election timing. CMS already announced mass cancellations, and the public still lacks a reliable breakdown of why coverage ended and how often the agency got it wrong. Asked for cancellation data, its press office told KFF Health News: “Currently, there is no additional data to share.”
Merritt also described an expedited reinstatement process. That is welcome, but the existence of a correction procedure does not tell us how quickly it works or how many people know they need it. CMS’s projected recovery and its cancellation total are not enough to evaluate the program. Wrongful losses and the time required to restore coverage belong in the accounting too.
Oz wants credit for protecting patients. The people his agency wrongly cut off have to count too.