AI tools fuel surge in Dutch insurance and retail fraud
Fraudulent insurance claims and online retail returns are rising sharply in the Netherlands, driven by AI tools that create realistic fake images and documents in seconds, De Telegraaf reports.
Large online retailers with automated return systems, including ASOS, Amazon, and Zalando, reported receiving at least 1,000 AI-manipulated claims in December alone. Some manipulations were detected, for example, through misaligned letters on garment labels, but many reportedly went unnoticed, leading to fraudulent payouts.
In one retail case, a customer purchased pants online for approximately 50 euros. Using ChatGPT, the customer manipulated a photo to make the pants appear severely damaged and sent it to the retailer with the claim, “My pants were delivered damaged.” Within seven days, the retailer either sent a replacement or refunded the purchase, even though the pants were undamaged.
The Dutch Association of Insurers highlighted a case in which a young man claimed to have lost 1,500-euro Cartier sunglasses borrowed from a friend. He submitted a fully AI-generated invoice and tried to claim the loss with two insurers. Both claims were rejected after verification through the CIS database, and the individual was registered in the fraud referral system.
Insurance fraud in the Netherlands rose to over 9,000 cases in 2024, up 1,000 from 2023, according to the Centrum Bestrijding Verzekeringscriminaliteit (CBV). The rise is linked to the widespread use of AI tools.
Stefan de Gooijer, an insurance expert at Pricewise, warned, “Take windshield damage, for example. An AI tool can create a crack that is nearly indistinguishable from a real one. Depending on the insurer, a damage expert may inspect the claim, or the payout may be issued based solely on a photo. This condition also applies to more severe damage. Everything is now so easy to manipulate that, if undetected, payouts will rise and premiums will increase. Every policyholder ends up paying for it.”
Patrick Petersen, a specialist in AI and retail, stated that fraudsters are now utilizing the most advanced image and video AI models, leading to a competition for control. The more advanced the techniques, the more complex and costly it becomes to prevent fraud. AI also increases the risk of wrongful suspicion, excluding honest customers from insurance. Human assessment aided by AI recognition tools remains essential.”
De Gooijer said even minor false claims are considered insurance fraud. Offenders can be listed in the CIS Extern Verwijzingsregister (EVR), face fines, criminal charges, or counterclaims, and risk cancellation of their current insurance. The registration lasts eight years, making it difficult to obtain new coverage.
