Insurers Say AI-Driven Billing Added $942 Million to US Healthcare Costs

Hospitals' use of artificial intelligence tools in insurance claims submissions generated an additional $942 million in healthcare spending over a two-year period, according to an analysis by the Blue Cross Blue Shield Association (BCBSA). That figure, released by one of the largest health insurance trade groups in the United States, places a hard dollar amount on a fight that has been building quietly between hospitals and insurers since AI tools entered the medical billing pipeline.

What the analysis found

The BCBSA analysis identified "a sharp increase in patients being documented as having complex conditions." Complex-condition coding typically triggers higher reimbursement rates, because payers compensate providers more for treating sicker patients. The insurers' association, however, argued there is a "clear disconnect between [medical] coding and treatment," stating that there is "no evidence of corresponding change in care delivered."

The distinction matters. If hospitals are coding patients as sicker without delivering more or different care, the added documentation translates directly into higher payments from insurers — and, ultimately, into higher premiums and spending across the system. The BCBSA's $942 million figure represents its estimate of that excess spending across the two-year window it examined.

The analysis, it should be noted, comes from the insurance side of a contested relationship. Hospitals and insurers have financial incentives that pull in opposite directions on billing questions. But the scale of the documented increase in complex-condition coding is the latest concrete data point in a broader argument about what happens when both sides of a trillion-dollar payment system deploy AI against each other.

The New York Times frames a wider trend

The New York Times flagged the BCBSA analysis as just the latest sign that AI is contributing to rising healthcare costs. The NYT's framing, reported alongside the analysis, makes a specific structural argument: battles between hospitals and insurers over treatments and payments are nothing new, but the use of AI on both sides appears to be making those battles worse.

The dynamic is straightforward to describe. Hospitals deploy AI tools to draft and optimize claims documentation. Insurers deploy AI tools to scrutinize, audit, and deny those claims. Each side's automation raises the stakes for the other side's automation. The result, in the NYT's telling, is an escalating contest in which the administrative machinery of healthcare expands and hardens — with costs flowing outward to employers, patients, and the system as a whole.

The $942 million figure gives that abstract contest a concrete anchor. It is the first major insurer-published estimate tying a specific dollar amount to AI-assisted claims documentation over a defined period.

A founder's warning: "bots fighting bots"

Not everyone inside the AI industry dismisses the concern. Dr. Shiv Rao, founder of AI startup Abridge, acknowledged that the spread of these tools could lead to "a horrible dystopic future nobody wants to live in," with "bots fighting bots, agents fighting agents."

Rao's language is unusually blunt for an executive whose company operates in this exact market. Abridge builds AI tools for clinical documentation — the category of software hospitals use to generate the records that feed into insurance claims. Rao has a direct commercial stake in the technology's adoption. His willingness to name a dystopic failure mode nonetheless signals that even suppliers of this software recognize the systemic risk the BCBSA is describing: an arms race in which automated documentation and automated review chase each other upward.

But Rao did not stop at the warning. He argued the same technology might also reduce tensions and cut costs. That is the industry's central bet: that AI-assisted documentation, done properly, reduces administrative friction, eliminates billing disputes rooted in poor records, and speeds payment — rather than inflating claims.

The BCBSA analysis suggests that, at least over the two-year period it measured, the inflation effect is winning. Whether Rao's cost-reduction scenario can materialize likely depends on how coding practices, payer audits, and regulator scrutiny evolve from here.

"Not a war" — "a one-sided blood bath"

The sharpest language in the dispute came from inside the insurance association itself. Luke Chalker, the BCBSA's senior vice president, resisted characterizing the situation as a battle between equals. "It's not a war," Chalker said. "It's a completely one-sided blood bath," with insurers on the losing side.

That framing is deliberately provocative, and it comes from a paid advocate for insurers. But it captures the asymmetry the BCBSA claims its data shows: hospitals control the documentation, hospitals deploy the AI that writes the documentation, and insurers must then pay against coding they cannot independently verify reflects delivered care. In the association's view, payers are absorbing automated documentation-driven cost increases with no counter-leverage of comparable scale.

Whether the situation is truly one-sided is contestable. The New York Times' reporting — which the BCBSA analysis itself surfaced — describes AI use on both sides of the fight, suggesting insurers are not passive participants. Hospitals, for their part, have long argued that insurers use prior authorization and automated review to delay or deny legitimate claims. Chalker's "blood bath" line is a rhetorical position in that larger contest, not a neutral finding.

Why this matters beyond the $942 million

The stakes extend well past a single dollar figure. Healthcare billing in the United States runs through diagnosis and procedure codes that determine reimbursement. When AI tools systematically shift documentation toward higher-paying code categories — whether intentionally, through design, or simply as a byproduct of thorough automated note-taking — the effect compounds across millions of claims. A two-year, $942 million measured impact in documented complex conditions is the kind of number that attracts the attention of regulators, congressional staff, and state insurance commissioners who oversee both payer solvency and provider billing conduct.

The episode also lands at a moment of intense scrutiny of AI in healthcare more broadly. Insurers themselves have faced lawsuits and public criticism over their own use of AI in claims decisions. Hospitals, meanwhile, are racing to adopt ambient documentation tools from vendors including Rao's Abridge and its competitors. The BCBSA analysis effectively argues that the hospitals' side of that adoption wave is already producing measurable cost inflation — an argument insurers now hold up against both providers and, implicitly, the AI vendors supplying them.

The core evidentiary claim — coding up, care flat — is the one hospitals and AI vendors will have to answer. If documentation of complex conditions rises sharply while delivered care stays constant, the question becomes whether AI-assisted records are capturing real patient complexity that human coders previously missed, or whether the tools are optimizing documentation for revenue. The BCBSA asserts the latter. Providers are likely to argue the former, or at minimum that the insurers' measurement of "care delivered" is incomplete.

What comes next

The BCBSA has now put a number and a named senior executive behind the claim that AI-assisted billing is inflating costs. Expect the association to use the analysis in negotiations with hospital systems, in public advocacy, and potentially in appeals to regulators. Expect hospital groups and AI documentation vendors to contest the methodology and the "blood bath" framing.

Rao's "bots fighting bots, agents fighting agents" scenario is the warning both sides now cite. The open question — and the one the next round of data will answer — is whether AI in medical billing converges on the cost-cutting outcome Rao predicts, or escalates into exactly the automated documentation arms race the insurers say has already added nearly a billion dollars to American healthcare spending in two years.