Bots Fighting Bots: Hospital AI Billing Added $942M in Costs

Hospitals have a new way to bill you — and it runs on AI. According to a new analysis from the Blue Cross Blue Shield Association (BCBSA), the hospital AI billing boom has already added about $942 million to what Blue plans paid out over two years. AI-assisted medical coding tools documented patients as sicker. But according to the insurers, the care they actually received didn't change at all.

The findings come from a three-page white paper, "Hospital Coding Intensity Analysis: Major Bowel Procedures," published this week by BCBSA researchers Chris Birkmeyer, David Wennberg, Keith Kamons and Luke Chalker, according to Awesome Agents. The team examined inpatient claims from the first quarter of 2023 through the fourth quarter of 2025, drawn from Blue Cross plans that together cover roughly one in three Americans. The association itself is a network of 31 independent insurers serving more than 100 million people, Reuters reported via SRN News. The story was first picked up on September 26 by TechCrunch.

The core complaint is what the association calls a "clear disconnect between coding and treatment." There is, in its words, "no evidence of corresponding change in care delivered." The patients didn't get sicker. Their paperwork did. It is the first hard dollar figure attached to hospital AI billing — and a preview of what happens when both sides of a market automate at once.

How AI turns medical charts into higher bills

Hospitals are plugging AI coding tools into their billing departments to catch conditions that human coders might miss. Some systems passively listen to patient conversations — so-called ambient scribes — while others scan existing medical records for billable conditions, according to The Nano AI, which cites Reuters and Fierce Healthcare.

The trend shows up clearly in the numbers. The share of inpatient cases billed to Blue plans as medically complex rose from 37% at the start of 2023 to 40% by the end of 2025, the analysis found.

In major bowel procedures, the shift was even steeper. Claims coded at the highest level of complexity rose from 10.2% to 22.7%, while non-complex cases fell from 36.6% to 32.8%. Secondary conditions spiked too: for people undergoing major bowel surgeries, partial blockages in the intestines rose 55% and acid overload increased 33% between early 2023 and late 2025. None of it was matched by a change in the treatment delivered.

About $653 million of the extra spending — roughly 70% of the total — came from secondary diagnoses, the additional conditions listed on a claim beyond the main reason for admission. A single abnormal lab value, the kind of detail an AI system flags automatically and a human coder might have left off the chart, could push a case into a higher-paying billing group.

That mechanism pushed more than 55,000 cases into higher-paying categories, at roughly $11,000 per case, according to the analysis.

One example stood out. In major bowel procedures, top-quartile hospitals diagnosed anemia at a rate 38% higher than their peers — yet transfused fewer of those patients, 16.9% against 19.3%. "If it was worth coding, there should have been something done," Razia Hashmi, the association's vice president of clinical affairs, said.

And the trend has room to run. More than 60% of US hospital systems are now using AI-based coding technology, according to Digital Today. A June survey cited by Fierce Healthcare found more than 63% of healthcare organizations use AI in their revenue cycle work.

Luke Chalker, the BCBSA's senior vice president of product and data science, put the conclusion bluntly: "AI is identifying more billable conditions, not sicker patients."

When both sides deploy the machines

Hospitals aren't the only ones automating. Insurers also deploy AI to scrutinize claims and decide whether treatments and payments are justified. The result, TechCrunch reports citing The New York Times, is that AI on both sides seems to be making long-running disputes between providers and payers worse — not better. Health insurers such as Centene have said the use of AI tools by health systems has led to aggressive or inappropriate reimbursement payments, according to the same Reuters wire report.

Dr. Shiv Rao, founder of AI healthcare startup Abridge, acknowledged the risk in stark terms. Unchecked automation on both sides, he said, could lead to "a horrible dystopic future nobody wants to live in," with "bots fighting bots, agents fighting agents." He added that the technology could instead reduce administrative friction and cut costs — if deployed differently.

For now, insurers say they're the ones losing. "It's not a war," said Chalker. "It's a completely one-sided blood bath."

This is one of the first places where autonomous systems on both sides of a real market are optimizing against each other — with a quantified cost landing on everyone else. I've been tracking a similar pattern in the AI agent world: autonomous agents escaping their intended scope, and the governance gap when enterprises can't even inventory the AI agents running inside their walls. Hospital AI billing is the same story wearing a white coat: automation behaving exactly as designed, at a scale nobody can fully audit.

The insurers are careful to note what the analysis can't prove. The data alone makes it difficult to conclude that AI directly caused the higher costs, as industry coverage has noted. What the analysis does establish is a sharp, AI-coinciding rise in billing complexity with no matching change in care — and a plausible mechanism for how it happened.

What happens next matters well beyond hospital finance departments. Higher billing intensity feeds premiums, employer costs and, eventually, what patients pay out of pocket. Payers are pushing back with their own AI, and regulators are watching. The question is whether the industry builds guardrails for automation-versus-automation — or lets the bots keep fighting it out on our bills.