Hunton Insurance Recovery Blog

They say lightning never strikes the same place twice, but for two policyholders who were burglarized in back-to-back years, it did, and a simple paperwork mix-up may have cost them their insurance coverage.  Their complaint alleges that Lemonade, their insurer, unfairly denied claims from a 2025 burglary because receipts that had been submitted with a 2024 burglary claim were inadvertently submitted again.  

The complaint raises interesting questions about how insurers detect fraud and the appropriate approach when discrepancies are identified.  What responsibility do insurers have to further investigate and allow the policyholder to explain?  These questions will become increasingly important the more that parties rely on AI in connection with insurance claims, increasing the likelihood that potential errors blossom into coverage disputes.

Background

In a recent complaint, two unlucky co-residents allege that Lemonade unreasonably denied coverage for losses stemming from a 2025 burglary.  Before the 2025 burglary, the co-residents were burglarized in 2024 at a different residence.  After the 2024 break-in, the residents filed claims with Lemonade for the stolen property, including three specific designer handbags.  Lemonade investigated and provided coverage for the 2024 burglary claims.

In 2025, the pair suffered their second burglary.  As before, the co-residents submitted claims to Lemonade under their separate insurance policies.  More than 90 items of high-end jewelry, watches, designer handbags, and other valuable property were submitted in the claims. 

Unfortunately, among the items submitted in 2025 were the same three designer handbags that had been submitted in 2024.  The complaint explains that the co-residents frequently buy designer merchandise and maintain a substantial collection of designer items like the ones stolen in the burglaries, leading to the mix-up. 

Lemonade smelled a rat.  It denied the co-residents’ claims on the grounds that one of them had materially misrepresented her claim by submitting documentation previously used in support of a prior claim and declared coverage void under both policies. 

So followed a lawsuit.  The complaint alleges that Lemonade unreasonably denied coverage based on an inadvertent error, which did not qualify as fraud, nor gave the insureds a chance to correct the error before denying coverage.  The co-residents also fault Lemonade for treating the separate claims as “one differentiated claim” and attributing one resident’s conduct to the other. 

Can AI tell fraud from mistake?

The complaint raises important issues about the grounds to deny coverage based on potential fraud, particularly as more and more insurers are using AI to handle claims.

First, this complaint reinforces that insurers will carefully review submissions and, if they suspect fraud, may deny the entire claim.  This is not the first time that insureds got in hot water after submitting claims for burgled designer goods:  a Real Housewife and her husband were famously charged with insurance fraud for designer goods claimed to have been lost in a home invasion.  Policyholders should expect insurers to review receipts and double-check all documentation, especially for high-value property.

Second, to err is human, and errors can be inadvertent; thus, insurers should provide an opportunity to explain.  Here, where two burglaries happened in short proximity, the insureds may have mixed up their receipts and confused which property was stolen when.  Regardless, insurers should have processes in place to detect potential mistakes and allow explanation.  Insurers that do not reasonably investigate alleged discrepancies do so at their own peril.

Third, policyholders should consider the use of AI in handling insurance claims and potential ramifications.  Although the complaint does not allege that the coverage decision was made by an AI tool, it highlights the types of discrepancies that automated fraud-detection tools are designed to identify.  Lemonade, for example, is “built on artificial intelligence” and as of 2025, about 55 percent of its claims are automated. 

While handling claims with AI may save money, it raises concerns that claims will be denied without fully accounting for context that a human reviewer would find important.  Researchers at Stanford, for example, noted that when “insurance staff don’t understand how AI works . . . opaque decision-making could lead to AI making mistakes or causing unfair outcomes.”  Lemonade itself has expressed concern about black-box algorithms (though it believes those concerns can be alleviated).  At least two federal courts have permitted discovery into an insurer’s use of AI in handling claims, suggesting that judges too are wrestling with these issues. 

Conclusion

The line between fraud and a simple mistake can be tricky to navigate.  Policyholders should double- and triple-check claim documentation to avoid unnecessary claim denials.  But insurers likewise have a responsibility to reasonably investigate discrepancies.  Use of AI in claims handling heightens these concerns.