Did Hospital AI Inflate the Bills? US Insurers Say $942 Million in Extra Payouts
A US insurer group analyzed claims data and estimates $942 million in extra spending over roughly two years. Hospitals have countered that shifts in patient acuity also need to be taken into account. Whether AI directly drove up costs remains unproven.
The Blue Cross Blue Shield Association (BCBSA), a group representing US health insurers, released an analysis on September 24 (local time) linking hospitals’ use of AI to rising medical costs. The findings come from a review of de-identified claims data submitted by member insurers. According to the group, from 2023 through 2025 there was an increase in claims that classified inpatients as more complex cases. BCBSA estimates the resulting additional spending at $942 million.
At the center of the dispute are the AI tools hospitals use when they bill insurers. These are not tools that decide what treatment to recommend for a patient. They are tools that help determine which diagnoses get recorded for a given episode of care, and how much gets billed. The crux of the issue is that for the same hospital stay, the amount an insurer pays can change depending on which diagnoses end up in the record.
Does One More Diagnosis Mean a Bigger Bill?
According to BCBSA, these AI tools comb through test results and clinical notes to surface diagnoses beyond the primary reason for admission. In effect, they help hospitals translate clinical documentation into billing codes. When an additional diagnosis is attached, that admission can be shifted into a category that pays more.
The association says that more than $650 million, or roughly 70 percent of its estimate, is tied to these added diagnoses. It also reported that more than 60 percent of hospital systems use such AI tools. What insurers object to is not the use of the tools themselves. It is whether the growth in recorded diagnoses was matched by a corresponding increase in actual treatment.
What Insurers and Hospitals Are Arguing About
BCBSA contends it found no evidence that treatment increased in step with the rise in diagnosis codes. Back in March, the group raised the same concern about claims for childbirth-related admissions. This time it widened the scope of its analysis to include major bowel surgery cases. That said, the $942 million figure is an estimate calculated from claims data. The number alone cannot establish that any individual diagnosis was improper, or that AI directly drove spending higher.
The American Hospital Association (AHA) pushed back on the insurers’ claims in material it published in August. Its explanation: the condition of inpatients themselves has changed, and care for patients with milder symptoms has shifted to outpatient settings. The AHA also stressed that clinicians review the documentation and that hospitals are obligated to follow billing rules. That rebuttal was not a response to the September analysis; it targeted earlier insurer claims.
Does Any of This Apply to Hospital Bills in Korea?
The analysis looks at US insurance claims data. There is no basis for applying the additional spending figure directly to Korea’s national health insurance system. It is also worth separating two distinct things: additional spending as calculated from insurance claims, and the out-of-pocket costs an individual patient pays. For readers in Korea, this is not a number to use in gauging changes in their own hospital bills or premiums.
The dispute puts two questions side by side. One is AI’s ability to surface more diagnoses from clinical records. The other is whether paying more money because of that is appropriate. What to watch going forward is how accurately those added diagnoses reflected patients’ actual conditions and the care they received. Whether AI use translates directly into higher spending, and whether patient costs rose along with it, will require separate verification.