Blog

The Takeaway

Insurance carriers should build strong databases to identify and flag fraudulent or AI-generated content. These tools can help carriers, coverage counsel, and others fight AI-driven fraudulent claims. Without such safeguards, the growing use of AI could cause insurance fraud to skyrocket.

Introduction

Revolutionary technology can carry significant consequences. Artificial Intelligence (AI) is no exception.

Before AI, fraudsters needed sophisticated technical skills to execute their schemes. It was common to see photoshopped receipts, doctored videos, and other manipulated “evidence” used to support a deceptive claim. Now, swindlers no longer need sophisticated technical skills. With a few keystrokes in a generative AI system, fraudsters can generate professional-quality fake receipts, videos, photographs, or audio recordings.

The Rise in Insurance Fraud

Property and casualty insurance fraud costs an estimated $45 billion annually (including home, auto, and business insurance). In addition, between 10% and 20% of insurance claims are fraudulent.[i] As a result, the FBI estimates that American families pay an extra $400 to $700 per year to cover these costs.

Technology is a powerful tool that helps carriers combat fraudulent claims. In 2021, a biennial study by the Coalition Against Insurance Fraud found that 80% of carriers used predictive modeling to detect fraud. This was a 55% increase from 2018. The Coalition also found that in the following two years, 21% of insurance companies had planned to invest in AI technology.

While insurance carriers increasingly use AI to prevent fraud, fraudsters increasingly rely on it to perpetuate their schemes. One survey[ii] found that 19% of UK claims handlers reported that 1 in 4 claims used fake supporting documents created or altered by AI and digital tools. Allianz, an insurance company, stated it’s seen a 300% increase in cases where apps were used to distort real-life images, videos, and documents.

Creation of an Insurance Carrier AI Database

Insurance carriers should follow other sectors and invest in AI fraud detection tools. For example, educators use AI detection software to combat student plagiarism and cheating. These software platforms detect AI-generated language, track writing patterns, and analyze metadata to confirm the authenticity of a document. Similarly, the financial sector uses AI to identify organized fraud and detect financial fraud patterns.

Insurance carriers should use AI tools to:

• flag and detect suspicious or AI-padded claims
• monitor historical data to recognize patterns of fraudulent claim attempts
• track repeat submissions from identified or suspected bad actors, including estimators, appraisers, physicians, and contractors

AI-assisted databases give insurance carriers’ special investigation units a powerful tool to detect and combat AI-enhanced fraud. Using AI image analytics, natural language processing by insurers, or AI-supported predictive models, carriers can also gather additional facts and data to more broadly support certain policy exclusions and defenses. For example, concealment or fraud provisions become easier to invoke when a fraudster uses AI to pad or alter a claim.

Increased AI use helps insurers transition from rules-based models (where insurers must hard-code software to flag what predicts potential fraud) to AI machine learning software (which discovers fraudulent patterns on its own). However, each model has trade-offs.

• Rules-based software relies on an insurer’s knowledge, limiting industry-wide detection.
• Machine-learning software might require third party software using large amounts of data from multiple insurers to identify patterns across insurers. Insurers providing outside companies with information, including sensitive or HIPAA-protected data, also poses a potential obstacle.

Regardless of the obstacles, Deloitte predicts that AI-driven fraud detection technologies could save insurers up to $160 million through 2032.

Concealment & Fraud: Application of an Insurance Carrier AI Database

Let’s look at an example of how this AI database could help.

In Illinois, insurers bear the burden of proof for fraud. They must show an insured intentionally concealed and/or misrepresented material facts about a claim. However, an insurer can do so using a preponderance of circumstantial evidence. Moore v. Farmers Insurance Exchange, 444 N.E.2d 220 (1982).

Under Illinois law, an insurer must prove that an insured intentionally misrepresented or concealed material facts. Traditionally, intent is a question for a jury. But AI-generated materials—such as fake photos or altered receipts—is an overt, intentional act designed to conceal or misrepresent the facts of a claim. Those actions can make intent easier to establish since their creation typically requires deliberate and intentional digital manipulation. That could allow insurers to resolve fraud claims earlier through summary judgment, reducing litigation costs.

Conclusion

The law on the use of AI is still developing, including its impact on concealment and fraud provisions. Courts have generally viewed AI use in litigation skeptically. (We’ve all heard accounts of the judiciary suspending, sanctioning, and admonishing counsel and pro se litigants for utilizing AI in court filings.) It’s not a big leap, then, to assume courts will apply the same sentiment to fraudster policyholders who utilize AI.

[i] Colorado State University Global’s White Collar Crime Research Task Force

[ii] Censuswide for Spout.ai