When it comes to attempts at insurance fraud, the story behind claims could be a major threat.
In a 2025 issue brief shared with Becker’s, the National Health Care Anti-Fraud Association outlined several possibilities for fraud that AI could spur: greater frequency of existing fraud behavior, false claims, identity theft, falsified authorization requests and appeals, and robocalls.
But Kurt Spear, Highmark Blue Cross Blue Shield’s vice president of financial investigation and provider review, believes fabricated medical records — created with the help of generative AI — are one of the most pressing issues as AI-driven fraud accelerates. The issue brief also said these records could be a concern.
“Historically, anti-fraud professionals would leverage that documentation to try to assess the validity of the claims,” he told Becker’s. “As we see that come in the door, we have to be more cautious and scrutinize those documents a lot more because they can be fabricated.”
Insurers can also leverage AI in determining fraud. AI use with claims more broadly has been a notable source of tension between payers and providers. Still, some AI-oriented use cases for combating fraud include data analysis, automated case generation and referral, automated text verifications for services and information gathering across sources.
When using tools to detect AI fraud, particularly with documentation review, Mr. Spear said they can compare lab reports and X-rays, determining whether AI was used and its legitimacy.
“Plans really have to have an AI strategy overall, meaning two things. [No. 1:] having an understanding and a framework for how it is going to be used against us, because it is being used against us,” he said. “[No. 2:] How can we leverage it for good? And I think if plans don’t think about that in those two buckets, and don’t have a strategy for it, then they are more susceptible to it, and they’re going to get hit by these fraudsters, and probably not even know about it until it’s too late.”
The issue brief also said AI could analyze policy loopholes and ambiguities, as well. Mr. Spear said fraud propelled by agentic AI, such as in that case, may not be as mature as that driven by generative AI. He said the industry finds it “a little bit harder to nail down” agentic AI fraud.
“Each agent might have a task that it performs, and one of those can be going out and scouring different publicly available medical policies, reimbursement policies, coding guidelines — and that’s how the bad actors can oftentimes find the loopholes,” he said. “It’s hard to determine if the actors are actually using agentic AI right now, or if they just know some of the areas we’re more susceptible to. … I wouldn’t be surprised at this point if it is being used, especially in some more of the complex scenarios. But I don’t think the agentic AI fraud is as mature as the generative AI fraud is.”
In his experience, fraud issues are typically not concentrated by specialty or provider type.
“It’s typically found more often in those out-of-area, out-of-network providers, versus a specific specialty or provider type,” he said. “The providers are harder to track down and identify because they’re not in network. They tend to open and close quickly.”
At the Becker's 5th Annual Fall Payer Issues Roundtable, taking place November 2–3 in Chicago, payer executives and healthcare leaders will come together to discuss value-based care, regulatory changes, cost management strategies and innovations shaping the future of payer-provider collaboration. Apply for complimentary registration now.