Artificial intelligence was supposed to make healthcare cheaper and smarter, but a new insurer-backed study says hospital billing bots are quietly inflating the bill by nearly a billion dollars in just two years.
The Blue Cross Blue Shield Association analyzed commercial hospital claims from early 2023 through the end of 2025 and found that AI-assisted coding tools helped push tens of thousands of cases into higher-paying categories without evidence that patients received more intensive care. According to the association, hospital use of AI in documentation and billing contributed an estimated $942 million in additional costs for Blue plans over 2024 and 2025 compared with a 2023 baseline, largely by increasing the number and severity of diagnoses attached to each stay. About $653 million of that total came from secondary diagnoses—conditions like anemia or low sodium that were added to claims and nudged more than 55,000 inpatient stays into more complex reimbursement tiers. On average, those extra-complex cases were paid nearly $11,000 to $12,000 more per stay, even though the underlying treatment patterns did not change in a meaningful way.
Earlier research from the association and its analytics partner Blue Health Intelligence suggested that the impact stretches well beyond the Blue Cross universe. That work estimated roughly $663 million in inpatient spending and at least $1.67 billion in outpatient spending nationwide could be tied to more aggressive, AI-enabled coding practices, pushing the potential cost impact of AI billing tools to around $2.3 billion. Within one participating Blue plan’s commercial inpatient population, per-member costs rose 9% between 2023 and 2024, with increased coding intensity—how complex patient records look on paper—accounting for about 20% of that jump. Insurers warn that those higher coded costs ultimately filter down into premiums and out-of-pocket bills for employers, patients, and taxpayers, even if the extra diagnoses never translate into additional care.
The report lands in the middle of what industry observers have dubbed an “AI bot war” between hospitals and insurers. Hospitals are increasingly using machine-learning tools to scan records, auto-generate documentation, and select billing codes that maximize legitimate reimbursement in a sprawling, rule-heavy system. Insurers, meanwhile, are deploying their own algorithms to flag potential upcoding, deny or adjust claims, and predict where costs might spike next. A March analysis found that healthcare AI spending hit about $1.4 billion in 2025, nearly triple 2024 levels, underscoring just how fast both sides are ramping up automated tools in the struggle over charges versus payments. Hospital groups argue that they need advanced software to keep up with complex payer rules and reduce administrative drudgery for clinicians, while insurers say some of these tools are being tuned to game reimbursement rather than reflect clinical reality.
Crucially, the Blue Cross studies focus on billing behavior, not whether AI is improving medical outcomes. A 2025 systematic review in npj Digital Medicine looked at 19 clinical AI interventions—things like diagnostic support tools and imaging algorithms—and found that many were associated with better diagnostic accuracy, gains in quality-adjusted life years, and meaningful cost savings compared to conventional care. That contrast highlights a split reality: AI aimed at clinical decision-making may often be cost-effective or even cost-reducing, while AI aimed at administrative coding can drive spending higher simply by changing what gets written on the chart. The Blue Cross Association itself notes that more severe billing codes are not necessarily tied to more procedures or longer stays, raising concerns that financial incentives are being reshaped without a corresponding benefit for patients.
For people on the receiving end of those medical bills, the immediate takeaway is less “robot doctors” and more “robot paperwork” quietly changing what care costs. If AI-driven coding keeps ratcheting up the apparent complexity of hospital cases, insurers say they will face pressure to increase premiums or tighten coverage, and regulators may be forced to scrutinize how far automated documentation tools can go before they count as abusive upcoding. Consumer advocates are already calling for more transparency around how AI is used in both billing and claims review, along with guardrails to ensure the technology serves patient care rather than revenue optimization. For a geek-culture audience used to dystopian sci-fi about runaway algorithms, the emerging real-world plot twist is subtler but no less unsettling: the bots aren’t overthrowing the system—they’re quietly rewriting the bill.
