The Blue Cross Blue Shield Association released a report Thursday saying hospitals' AI coding tools added $942 million to what its 31 member plans paid for inpatient care in 2024 and 2025, compared with 2023. The association says $653 million of that came from hospitals billing more often for secondary conditions, the extra diagnoses that push a stay into a higher-paying billing tier. The share of inpatient claims coded as medically complex rose from about 37% in early 2023 to about 40% by late 2025. The tools include software that scans patient records for conditions to code and "ambient scribes" that listen to doctor visits and draft the notes. The report came out a day after Dr. Mehmet Oz, who runs the Centers for Medicare and Medicaid Services, told an Oracle health conference in Orlando that AI will make billing more expensive before it makes it cheaper.

1. Insurers Say the Diagnoses Went Up and the Treatment Didn't (BCBSA's Luke Chalker and Razia Hashmi, Centene's Sarah London)

Insurers say the diagnoses went up and the treatment didn't, and that's how they know.

The diagnoses rose. The treatment didn't. That's the whole argument. Luke Chalker, the Blue Cross Blue Shield Association's senior vice president of product and data science, co-wrote the report. He puts it this way: "If patients are truly sicker, we'd expect to see more treatment." Instead, he says the "disconnect between diagnoses and treatment suggests that AI is identifying more billable conditions, not sicker patients."

Anemia is the clearest example. The association flagged the same pattern in maternity admissions back in March. Dr. Razia Hashmi, the association's vice president of clinical affairs, says patients newly diagnosed with anemia did not get more blood transfusions, the standard treatment. For patients having major bowel surgery, secondary diagnoses of partial intestinal blockages rose 55% and excess acid in the body rose 33% between early 2023 and late 2025. Each case that tipped into the complex tier paid roughly $12,000 more.

Other insurers have been saying the same thing for a year. Centene CEO Sarah London described "pockets where folks coming into the emergency department with a fever, all of a sudden all have sepsis." Her CFO, Andrew Asher, admitted last fall that hospitals "have gotten better organized around the application of AI for coding than payers. But we're going to catch up to that."

They reject the idea that this is an even fight. Chalker told the New York Times it is "not a war" but a "completely one-sided bloodbath." Blue Cross doesn't use AI to process claims, he said, only to guard against billing it considers inappropriate.

They want rules. The association wants clear expectations for hospitals that use AI coding tools, so billing matches the care delivered. It also wants the payment tiers changed so an extra diagnosis stops paying so well.

2. Patients Really Are Sicker, and Insurers Are the Upcoders (The AHA's Aaron Wesolowski, McLaren CFO Dave Mazurkiewicz, HCA CFO Michael Marks)

Hospitals say the insurers' math skips the obvious explanation, and that the insurers are the ones with an upcoding record.

Hospitals say the patients just got sicker. The American Hospital Association's fact sheet on the issue says hospitals "are increasingly caring for patients with higher acuity, a trend that is appropriately reflected in provider coding practices." It points to an aging population, more chronic disease, and a case-mix index up about 5% from 2019 to 2024. Aaron Wesolowski, the AHA's vice president of research strategy and policy communications, says easier cases have moved out of the hospital, so the care left inside "is naturally higher acuity."

Writing more of it down isn't fraud. Dr. Jared Dashevsky, an internal medicine physician who writes the Healthcare Huddle newsletter, makes that argument: "A tool that captures everything clinically relevant will naturally generate more complete, higher-acuity documentation." Ambient scribes hear the whole visit, he says, and catch conditions a rushed doctor used to leave out.

The hospitals say they're reacting to denials. Dave Mazurkiewicz, chief financial officer of McLaren Health Care in Michigan, told the New York Times, "All the insurers are using A.I. to scan our charts to look for claims to deny. For the same reason, we're looking at the same charts today." HCA Healthcare CFO Michael Marks called his system's AI tools a response "to the growing denial and underpayment activities from the payers." Five HCA-affiliated Texas hospitals sued Independence Blue Cross earlier this month over denied claims.

And the insurers are the ones with upcoding cases on the record. The AHA's fact sheet lists three. The Medicare Payment Advisory Commission, which advises Congress, reported in 2025 that upcoding contributed to $40 billion in overpayments to Medicare Advantage plans. One insurer settled a Justice Department upcoding case in March. Massachusetts sued another in May. On September 8, Inside Health Policy reported that the association had asked Congress and the Trump administration to stop insurers from falsely accusing providers of using AI to inflate their coding.

3. Everyone Knows AI Inflates Bills. Blame How Hospitals Get Paid (CMS Administrator Mehmet Oz, PHTI's Caroline Pearson, Duke-Margolis's Christina Silcox)

Regulators and analysts say both sides are right about each other. They want to change what gets paid for.

Medicare's own chief says the insurers have a point, for now. Speaking at Oracle's health summit on Wednesday, Oz said, "Short term, AI is going to be inflationary because it's going to turbocharge the ability of the current billing systems to work more effectively." He wants more providers in accountable care organizations, where they get paid for outcomes instead of for each service.

In private, nobody disputes the effect. Caroline Pearson, executive director of the Peterson Health Technology Institute, described a roundtable her group held with insurers, hospitals, and investors: "The investors, the health plans, and the providers, in private, were like, 'OK, well, it's quite clear scribes are increasing coding intensity. One hundred percent.'"

Nobody wins when AI fights AI. Christina Silcox, research director of digital health at Duke's Margolis Institute for Health Policy, put it that way: "The idea of (AI) bot versus bot is intrinsically a situation where no one's going to win." Even a scribe vendor agrees. Dr. Shiv Rao, founder of Abridge, told the Times, "That's the danger, where it's bots fighting bots, agents fighting agents, a horrible dystopian future nobody wants to live in."

The billing tiers reward every extra diagnosis. For major bowel surgery, the report's own example, Medicare's payment weights show what a tier jump is worth: one qualifying secondary diagnosis raises the payment about 42%, and a major complication raises it about 173%.

Where This Lands

The insurers say the numbers prove hospitals' AI is finding billable diagnoses. They want rules on the tools and a rewrite of the billing tiers. The hospitals say their patients are sicker and their software just writes down what doctors used to miss. They also say the insurers accusing them have their own upcoding settlements to answer for. Oz and the analysts say both are right about each other, and that AI on both sides will keep finding diagnoses as long as the billing system pays for them. That leaves Congress and CMS. Either they write rules for the coding tools, or both sides keep fighting it out claim by claim.

Sources