No-claims was one of four tests Oz cited, and those enrollees paid no premium

The posts

Rupar’s post accompanies C-SPAN video of a 22 September news conference in Washington, where Vice President JD Vance, who chairs the White House Task Force to Eliminate Fraud, and Oz, administrator of the Centers for Medicare & Medicaid Services, announced the cancellations.^[12][1][2] The reply appeared alongside it in the screenshot the reader sent; we check its premise here as the reader asked.

Verdicts at a glance

• Mixed

  The administration will kick more than a million people off health insurance

  About 760,000 have been canceled. Roughly 419,000 more are under review, not automatically cut. Together they come to about 1.18 million, but only if every case under review ends in cancellation.

• Misleading

  …because some of them have “never filed a claim”

  Oz did cite never filing a claim, but as one of four combined tests. CMS’s own release describes the cancellations as unauthorized enrollments confirmed with insurers and does not mention claims at all.

• False

  Reply: it is now “fraud” to pay for health insurance you don’t use

  The flagged enrollees had their entire premium covered by federal tax credits. Someone paying their own premium and never filing a claim would not meet the tests Oz described.

• Contested

  Oz: zero-claim enrollment points to fraud

  The share of marketplace enrollees with no claims rose from 19% in 2021 to 35% in 2024. Paragon Health Institute calls that evidence of fraud. Insurers’ trade group AHIP and KFF say healthy, younger or part-year enrollees explain much of it.

• Unproven

  Implied: all 760,000 were fraudulent or ineligible

  CMS says each cancellation was confirmed as unauthorized with insurers but has not published its method. KFF says many may have been enrolled without their knowledge, and questions whether all were.

760,000 canceled, about 419,000 under review

CMS said it canceled “approximately 315,000 unauthorized enrollments covering more than 760,000 individuals,” expected to return about $2.2 billion in taxpayer-funded subsidies.^[3] An enrollment can cover a whole household, which is why the two numbers differ. Reuters reported that Vance said the administration would further verify “between 419,000 and 450,000 more people”; NBC News and ABC News reported 419,000.^[4][5][6] IBTimes, citing CMS, reported those cases were “undergoing additional verification rather than being automatically cancelled.”^[7]

So “more than a million” describes a ceiling, 760,000 plus the review group, about 1.18 to 1.21 million. It becomes true only if nearly everyone in review is removed.

The four tests

According to NBC News, Oz said the administration looked at “a combination of factors it considered suspicious.” The people had:^[4]

1. been enrolled with the help of a broker or agent;
2. never filed a claim;
3. had their entire monthly premium covered by federal tax credits;
4. had no Social Security number or immigration documents CMS could verify in real time.

Reuters added that Oz said the group was treated as “phantoms” because they had “never filed a claim” and were “unreachable despite repeated contact attempts.”^[5] “We call them phantoms, because we believe most of them don’t exist or they had no idea they had coverage,” Oz said.^[4]

CMS’s written release does not mention claims. It describes “unauthorized enrollments,” promises to hold “agents and brokers accountable when they break the rules,” and says CMS will work with insurers to “cancel those confirmed to be unauthorized.”^[3] Vance described the canceled group as “a mix of both phantom people, but also real people who just don’t meet the eligibility requirements,” including people signed up by brokers without their knowledge.^[5]

Rupar’s post is accurate that Oz invoked never filing a claim. It leaves out that no-claims was paired with a fully subsidized premium and an unverifiable identity, which changes what the policy targets.

Does “no claims” signal fraud?

At the briefing, CNBC reported, Oz said around 35% of people in the program “have never used the program,” meaning no prescription and no doctor visit.^[2] It echoes an analysis by the Paragon Health Institute, the think tank of Brian Blase, a health adviser in Trump’s first administration.^[1]

Compared against: zero-claim rates in the ACA marketplace in earlier years and in employer-sponsored insurance. Marketplace: 19% of enrollments had no claims in 2021 and 35% in 2024. Employer plans: an average of 23% of enrollees did not use their insurance each year from 2018 to 2022.^[8]

The case that it signals fraud

Paragon argues that “rising numbers of zero-claim enrollees” bolster “evidence of rampant fraud,” that the COVID-era credits created “zero-dollar plans that are fully subsidized by taxpayers and require no enrollee contributions,” and that the rise “is not explained by meaningfully higher enrollment among young, healthy beneficiaries.”^[9] A Government Accountability Office covert test found the federal marketplace approved subsidized coverage for almost all of its 24 fictitious applicants in 2024 and 2025.^[10] Broker fraud is not disputed: the Biden administration suspended 850 brokers and agents suspected of fraud in 2024, though, CNN reports, Trump officials later quietly reinstated several hundred of them.^[1]

The case that it does not

AHIP, the trade group for the insurers themselves, calls the argument a myth: “A ‘no-claims’ year is evidence that a consumer stayed healthy or only had a few months of coverage,” and it is “commonplace for consumers to enroll in a health plan for only a portion of a year.”^[11] KFF’s Cynthia Cox told KFF Health News that the federal data can count enrollees twice if they switched plans, and that the subsidy expansion made the marketplace younger, and younger people use care less.^[8]

The four-test design partly answers this critique. Healthy people who pay part of their own premium, or whose identity checks out, would not be flagged. The critique still applies to people with a fully subsidized plan and missing documents who were real and eligible, and CMS has not said how many of those it found.

What the posts imply

Reply: it’s fraud to pay for insurance you don’t use (False)

The tests targeted plans whose entire premium was covered by federal tax credits, so the enrollees in question were not paying for their coverage.^[4] No official described not using paid-for insurance as fraud. The reply rests on a reading of Rupar’s post that the post itself invites: that no claims alone was the trigger.

Implied: all 760,000 were fraudulent or ineligible (Unproven)

CMS says the cancellations were confirmed as unauthorized with insurers.^[3] Oz himself said some of those under further review “had something about their story that was a little bit different than what we expected from the 100% fraud story.”^[4] KFF’s Cox said “it is entirely possible that many of these people were enrolled without their knowledge or were so-called phantom enrollees,” but also questioned “whether this was the appropriate process by which to identify fraudulent enrollees, and also whether all of them were indeed fraudulently enrolled.”^[4][10] Until CMS publishes how it confirmed each case, the share that was fraud cannot be checked.

Implications

For people who were real and eligible. A person enrolled by a broker, with a fully subsidized plan, who stayed healthy and whose documents did not verify in real time, meets every test Oz described. Anyone in that position should check their coverage status with HealthCare.gov. We did not find published guidance from CMS on how people who were canceled in error can be reinstated.

For the debate. Both viral framings miss. The post makes the policy sound as if it punishes healthy customers, and the reply builds on that. The administration’s framing, “phantoms” who “don’t exist,” claims more certainty than it has published evidence for. The accurate description sits between them: a screen for fully subsidized, broker-enrolled, unverifiable and unused plans, whose error rate is unknown.

What to watch. Whether CMS publishes its method and the outcome of the roughly 419,000 reviews; the agent and broker moratorium CMS announced for the 2027 plan year; and enrollment figures for 2027, which will show how many people return.^[3]