Published: September 27, 2026 · Last updated: September 27, 2026
TL;DR: The Blue Cross Blue Shield Association says AI powered medical coding tools drove $942 million in extra healthcare spending over two years, mostly by making patients look sicker on insurance paperwork than they actually are. Hospitals aren’t necessarily giving worse or better care, the AI is just describing that same care in more expensive terms. One health tech founder is already warning about “bots fighting bots.”
Nobody got a better diagnosis. Nobody got a new treatment. What changed is how it got written down, and that alone was enough to add nearly a billion dollars to what insurers paid out.
Same Care, Fancier Paperwork
The Blue Cross Blue Shield Association’s analysis is not subtle about what it found. Hospitals have been rolling out AI tools that help staff document patient conditions when submitting insurance claims, and those tools are very good at their job, arguably too good. BCBSA found a sharp increase in patients being coded as having complex conditions, the kind of documentation that triggers higher reimbursement rates. But when analysts looked for a corresponding change in the actual care those patients received, they found nothing. Same treatments, same outcomes, just a more expensive paper trail attached to them.
Over two years, that gap between documentation and delivered care added up to $942 million in additional spending, according to BCBSA’s figures. That number does not represent new medical value. It represents AI systems getting better at describing existing care in the specific language that insurance reimbursement formulas reward.
A Founder’s Warning About Bots Fighting Bots
Dr. Shiv Rao, who founded the AI medical documentation company Abridge, did not shy away from where this trend leads if nobody intervenes.
“A horrible dystopic future nobody wants to live in” is what Dr. Rao called the endpoint of this arms race, describing a system where “bots fighting bots, agents fighting agents” replaces actual clinical judgment on both sides of the claims process.
The mechanics are straightforward once you see them. Hospitals deploy AI to code claims in the most favorable, defensible way possible. Insurers deploy their own AI to review and challenge those claims. Neither side’s software is doing anything a human coder couldn’t theoretically also do, they are just doing it faster, at greater scale, and with none of the friction that used to slow down aggressive coding practices.
Luke Chalker, a senior vice president at BCBSA, put it more bluntly than most industry executives are willing to on the record. He described the current dynamic as anything but an even fight.
“It’s not a war. It’s a completely one sided blood bath,” Chalker said, characterizing insurers as the side losing ground in this particular AI arms race.
Dr. Rao offered a counterpoint worth sitting with too. He suggested that better AI on both sides could eventually reduce tensions rather than escalate them, if the technology matures past the current phase of exploiting reimbursement loopholes and into something closer to genuinely more efficient, accurate documentation. That is a real possibility. It is also exactly what every party in an escalating arms race says about their own side’s next upgrade.
Frequently Asked Questions
Does this mean AI is making healthcare worse for patients?
Not directly. BCBSA’s findings focus on billing and documentation, not clinical care quality. The concern is about where the cost of this documentation arms race eventually lands, which is typically premiums.
Who is Blue Cross Blue Shield Association here, an insurer or a neutral party?
BCBSA represents a network of insurance plans, so it has a direct financial interest in this issue. Its findings should be read alongside that context, even where the underlying documentation gap appears to be real.
What is Abridge, and why does its founder’s opinion matter here?
Abridge builds AI tools used in clinical documentation, putting Dr. Rao on the hospital side of this exact dynamic, which makes his warning about an escalating arms race notable coming from inside the industry building the tools.
Related: Google Just Put a Buy Button Inside Its Chatbot, and Only Flipkart Gets It and An SEO Pro Tracked Every Way Claude Wrecked His Work.
Bottom Line: This is not a story about AI diagnosing anyone better or worse. It is a story about two industries pointing increasingly sophisticated software at each other over a paperwork negotiation, and nearly a billion dollars sliding through the cracks in between. Somebody pays for that gap eventually, and it is rarely the hospital or the insurer.
Follow Teck Hustlers: Facebook · X · Instagram · LinkedIn · TikTok · Pinterest
