The Blue Cross Blue Shield Association said on October 1, 2026, that its member health plans identified approximately $942 million in questionable hospital charges between 2023 and 2025 that it attributes to AI-assisted medical coding tools, reigniting a long-running dispute between insurers and hospitals over how AI software is reshaping medical billing. The American Hospital Association pushed back immediately, calling the analysis incomplete and lacking the clinical context needed to judge whether the flagged charges were actually improper.
How AI Coding Tools Work
Hospitals have increasingly adopted AI-powered coding software that automatically scans clinical documentation and assigns billing codes, a process traditionally handled by trained human coders. Proponents say the technology speeds up reimbursement, reduces human error, and frees coding staff for more complex cases. Blue Cross Blue Shield’s analysis suggests the same automation may also be nudging codes toward higher-reimbursement categories more consistently than human coders did, a pattern sometimes called “upcoding” when it isn’t clinically justified.
The Insurer’s Case
Blue Cross Blue Shield’s member plans, which collectively cover more than 100 million Americans, say their internal audits found a statistically significant jump in high-complexity billing codes at hospitals that adopted AI coding tools compared with those that hadn’t, even after controlling for patient mix. The association argues this pattern, compounded across millions of claims, is what produced the nearly $1 billion figure, and it is pushing for tighter oversight of how coding software is validated before deployment.
Hospitals Push Back Hard
The American Hospital Association rejected the framing, arguing that more accurate, AI-assisted documentation of patient complexity is likely to produce legitimately higher codes in many cases, particularly for sicker patients whose conditions were previously undercoded by overworked human staff. The group also noted that insurers have a direct financial incentive to characterize more claims as questionable, since doing so can justify delaying or denying payment.
A Broader Pattern of AI-on-AI Friction
The dispute reflects a broader dynamic emerging across U.S. healthcare: insurers are deploying their own AI systems to review and flag claims, while hospitals use AI to generate them, creating a largely automated back-and-forth with limited human review on either side. Reports have found that a growing majority of health insurers now use AI for prior authorization decisions, including denials, a trend that has drawn criticism from patient advocates concerned about reduced human judgment in coverage decisions.
What Independent Experts Say
Health policy researchers note that neither hospitals’ nor insurers’ AI tools have been subject to much independent, peer-reviewed validation, making it difficult for outside observers to adjudicate competing claims about accuracy. Some researchers have called for a neutral, third-party auditing standard for both coding-generation and coding-review AI systems, arguing that without such a standard, disputes like this one will likely recur and escalate as both sides deploy more sophisticated tools.
How Other Insurers Are Responding
UnitedHealthcare, Cigna, and Aetna have not published similar figures publicly, but industry trade groups say most major insurers are running comparable internal audits of AI-assisted coding patterns at hospitals they contract with, suggesting Blue Cross Blue Shield’s disclosure may be the first of several similar reports to surface in coming months. Some insurers have begun requiring hospitals to disclose which AI coding vendors they use as a condition of contract renewal, a step toward the kind of oversight regime regulators have so far declined to mandate at the federal level.
What Regulators Have Said So Far
The Centers for Medicare and Medicaid Services has not issued specific guidance on AI-assisted coding accuracy, leaving the dispute largely to play out between private insurers and hospital systems without a clear federal referee. Some state insurance commissioners have begun fielding complaints from both sides, though none have opened formal investigations into AI coding practices as of this month, according to people familiar with state regulatory agendas.
What’s Next
Blue Cross Blue Shield said it plans to share its methodology with regulators and is pushing for stronger federal guidance on AI-assisted billing practices, while the American Hospital Association has called for a joint industry study rather than unilateral insurer audits. Given how much money is at stake industry-wide, the disagreement is likely to feature in upcoming congressional hearings on AI in healthcare, and could ultimately shape how aggressively hospitals continue to roll out AI coding tools. Congressional staff on relevant health committees have also requested briefings from both the insurer and hospital trade groups, according to people familiar with the outreach, signaling the dispute could surface in hearings before year’s end.
Photo: shameersrk / PIXABAY via Pixabay
