Blue Cross report links hospital AI to $1B rise in costs

Hospitals have been quietly upgrading their billing departments with AI tools that listen to doctor-patient conversations, auto-generate diagnosis codes, and optimize claims submissions. According to a new report from the Blue Cross Blue Shield Association, those tools are working exactly as designed. The problem is that “working as designed” appears to mean squeezing significantly more money out of every patient visit.

The BCBSA report, produced by its analytics arm Blue Health Intelligence, analyzed commercial inpatient claims covering approximately 62 million members over a three-year window ending in March 2025. The core finding: hospitals using advanced AI technologies like ambient listening and autonomous coding showed sharp increases in the number of diagnoses categorized as complex, driving up claims costs without a corresponding improvement in care quality.

The numbers behind the billing creep

At the top 10% of hospitals by coding growth, the share of inpatient admissions coded as complex jumped from 46.8% to 59.8%.

One of the most telling examples involves maternity care. At high-growth hospitals, diagnoses of acute posthemorrhagic anemia in maternity cases surged from roughly 4% to over 12%. Transfusion rates stayed flat, suggesting the AI tools were finding billable diagnoses that human coders had previously overlooked. That particular discrepancy alone contributed an estimated $22 million in questionable charges.

The report attributes roughly $663 million in excess inpatient spending to these AI-driven coding practices. Outpatient spending, where AI tools are also proliferating, added at least $1.67 billion more. Total additional costs from the phenomenon: approximately $2.3 billion nationwide.

Between 2023 and 2024, per-member inpatient costs within the studied BCBS population rose by 9%. The report estimates that coding intensity was responsible for about 20% of that increase.

How AI turned billing into an arms race

Hospitals argue the uptick in complex coding reflects more thorough documentation in accordance with evolving clinical guidelines, combined with genuinely sicker patient populations seeking care. But the BCBSA data suggests the acceleration is too steep and too concentrated among AI-adopting facilities to be explained by clinical factors alone.

By late 2026, nearly half of hospitals were utilizing AI for billing, coding, or claims processing. PwC’s 2027 forecast projects AI will contribute to an 8.5% to 9% increase in commercial medical costs.

Insurers push back, but the leverage is shifting

BCBSA and its member plans are responding with their own data-driven oversight tools, building detection models that flag anomalous coding patterns and targeting facilities where diagnosis complexity has outpaced measurable changes in patient acuity or treatment intensity.

Employer-sponsored health plans, which cover the majority of commercially insured Americans, ultimately absorb these cost increases through higher premiums. The Centers for Medicare and Medicaid Services has been monitoring AI’s role in clinical documentation and coding, though enforcement actions specifically targeting AI-driven upcoding remain limited.