Blue Cross Blue Shield Says Hospital AI Coding Added $942 Million to Costs
Blue Cross Blue Shield says hospital AI claim coding added $942 million over two years, with more complex diagnoses but no matching change in treatment.
Summary
The Blue Cross Blue Shield Association said on September 26, 2026, that hospitals using artificial intelligence to submit insurance claims generated an additional $942 million in healthcare spending over two years. Its analysis found a sharp increase in patients documented with complex conditions, but no evidence that delivered care changed accordingly, exposing a disconnect between medical coding and treatment.
AI use by hospitals and insurers may be intensifying longstanding disputes over treatments and payments. Luke Chalker, the association’s senior vice president, characterized the imbalance as a one-sided rout with insurers losing. Abridge founder Dr. Shiv Rao warned that automated agents could end up battling one another, although he said AI could instead reduce tensions and costs. Details on corrective policies, hospital responses and timing remain limited.
Positives
- Blue Cross Blue Shield Association quantified the disputed impact at $942 million over two years, providing a concrete measure of AI-assisted coding costs.
- Dr. Shiv Rao said AI could reduce payment tensions and lower healthcare costs despite current concerns.
- Blue Cross Blue Shield compared medical coding with delivered care, identifying a specific disconnect for hospitals and insurers to address.
Risks & concerns
- Hospital AI claim coding generated an additional $942 million in healthcare spending over two years, Blue Cross Blue Shield Association found.
- Patients were increasingly documented with complex conditions without evidence of a corresponding change in care delivered.
- AI use by both hospitals and insurers may intensify longstanding disputes over treatments and payments.
- Luke Chalker described insurers as losing a one-sided struggle against AI-assisted hospital claims.
- Dr. Shiv Rao warned that automated agents could enter a dystopian cycle of fighting other automated agents.