A new analysis from the Blue Cross Blue Shield Association (BCBSA) finds that hospitals’ growing use of artificial intelligence to assign medical billing codes added close to $1 billion in extra costs to Blue Cross Blue Shield health plans between 2023 and 2025.
According to the insurer’s trade group, AI-assisted coding helped drive $942 million in additional costs over that period, with much of the increase tied to secondary diagnoses that pushed patients into higher-paying reimbursement categories. Medical coders convert procedures and diagnoses into standardized codes used for insurance billing, and BCBSA says the rise in what it calls “complex coding” coincides with a stretch when roughly 60 percent of hospital systems adopted AI coding tools.
About 70 percent of that total, or $653 million, involved additional diagnoses that did not correspond to any change in a patient’s actual treatment, according to Luke Chalker, BCBSA’s senior vice president of product and data science. The report notes that many of these secondary diagnoses “may be derived from single laboratory values, making it particularly well suited for detection by AI tools,” and concludes that “there is a clear disconnect between coding and treatment.”
Chalker told CNBC that “while multiple factors contribute to coding intensity, the findings suggest AI-enabled coding and documentation tools are playing a role.” He said the added costs don’t stop with insurers. “Those costs can eventually show up in the form of higher premiums and out-of-pocket costs,” he said. Benefits consulting firm Marsh projects per-employee health coverage costs will rise 8.2 percent on average in 2027, which would be the steepest increase since 2003.
A Dispute Over Blame
Not everyone agrees AI is the root cause. Christopher Whaley, a health economist at Brown University who studies hospital coding, argues the technology isn’t creating new billing incentives so much as exploiting ones that already existed. AI is “accelerating, and in some sense making it easier to capture, the existing and underlying billing incentives that are in the system,” he said.
Whaley also cautioned against assuming the added diagnoses are improper. “In many cases, the diagnoses are legitimate and weren’t captured,” he said, though he acknowledged some flagged conditions “clinically just don’t really matter and don’t influence the patient’s care” even as they support higher bills.
The American Hospital Association rejected BCBSA’s framing outright. A spokesperson told CNBC that “patients today are older and more clinically complex” and that AI tools help providers “appropriately capture their patients’ conditions to aid in care planning.” The group said BCBSA’s analysis “lacks the context needed to meaningfully assess how these tools impact healthcare quality, patient access, or spending,” and turned the criticism back on insurers, calling it “particularly troubling to see insurers raising concerns about provider coding while continuing to rely on automated downcoding and denial practices that can impede coverage of medically necessary care, add burden on the workforce, and increase costs through administrative waste.”
Chalker acknowledged that Blue Cross Blue Shield companies also use AI in reviewing claims, but said human oversight remains part of the process. “Any clinical denial is always reviewed by a qualified human clinician,” he said.
An “Administrative Arms Race”
Whaley described the standoff between hospitals and insurers as an “administrative arms race,” with both sides spending heavily on technology that does nothing to improve patient care. “Whether it’s on the hospital side or the insurer side, these tools and technologies are both very expensive, and also have nothing to do with providing appropriate care to patients,” he said.
Marisa Greenwald, a partner at Oliver Wyman’s Health and Life Sciences practice who advises hospitals and insurers on AI adoption and revenue-cycle management, offered a more mixed take. She said AI can ease physicians’ workloads, giving them “some semblance of work-life balance back,” and that AI-assisted documentation can help providers “catch additional acuity and diagnosis components,” improving billing accuracy in some cases.
Still, she said separating genuine improvement from overreach is difficult. “It’s hard to disentangle how much of it is better accuracy,” she said. “There’s always going to be misuse and user error and overcoding.” She does not expect tensions between hospitals and insurers to ease. “The arms race is poised to exacerbate,” she said. “The hope is going to be that on both sides of the equation, the players recognize that all we’re doing is adding” costs.
