Insurance companies reported a record 1.93 billion CZK in attempted insurance fraud last year, according to data from the Czech Insurance Association (ČAP). This surge is driven by the use of generative artificial intelligence to manipulate photos and documents, creating highly credible forgeries that bypass traditional manual review processes.
How is AI changing the nature of insurance fraud?
Generative AI tools now allow fraudsters to create nearly perfect forgeries of invoices and medical reports, according to Veronika Hešíková, spokesperson for Uniqa insurance. In one case handled by Kooperativa, a claimant attempted to use an AI-modified invoice to cover a broken iPhone; the software had actually been used to alter a receipt for a MacBook purchased in 2008. These digital modifications often include subtle adjustments to metadata, lighting, and image structure that are difficult for human investigators to spot without specialized forensic software.
Insurance investigators are increasingly using “documentary forensic analysis” to catch fraud. Kooperativa’s systems alone scan approximately 1.1 million documents annually, automatically flagging inconsistencies that suggest tampering.
Why are fraud attempts rising despite better detection?
While insurers are catching more fraud, the total value of attempted claims continues to climb. Data from the Czech Insurance Association shows that the average value of a detected fraud attempt reached 265,000 CZK last year, an 11% increase compared to the previous period. Experts suggest that as companies deploy “technological screening,” fraudsters are testing the limits of these systems with high-stakes attempts. Michal Kusyn of the technology firm Trask notes that insurers must shift from testing AI tools to building “robust daily defenses” to maintain sector stability.
How do insurance companies fight back with technology?
Insurers are shifting their strategy from manual investigation to automated, algorithm-driven screening. Radek Juhaňák, director of liquidation at Allianz, states that the industry’s ability to detect fraud is evolving as quickly as the threats themselves. By utilizing scoring models and forensic image analysis, companies like Uniqa prevented approximately five million CZK in fraudulent payouts in the first quarter of this year alone. These systems allow investigators to compare current claims against historical data, such as a vehicle’s past repair records or a property’s original condition.
Comparison of Fraud Attempt Trends
| Year | Nominal Value (CZK) | Reálná hodnota (2025 prices) |
|---|---|---|
| 2025 | 1.930 mld. | 1.930 mld. |
| 2023 | 1.664 mld. | 1.747 mld. |
| 2021 | 1.249 mld. | 1.670 mld. |
Frequently Asked Questions
What is the most common type of insurance fraud today?
According to Dita Němečková of Kooperativa, insurers most frequently encounter manipulated medical records and falsified invoices for consumer goods.
Can AI be used to prove a claim is legitimate?
Yes. Insurers use the same forensic tools to verify the authenticity of documents and photos, which can speed up the payout process for honest clients by automating the verification phase.
Are insurance companies investigating fewer cases?
Yes, in terms of volume. Because automated systems block many fraudulent applications before they are processed, the number of manual investigations has dropped to roughly 9,000 cases annually, even as the value per case rises.
Subscribe to our weekly newsletter for the latest updates on digital security and insurance technology trends. Sign up here.
