Fraud Detection 2026 – Trends in Theses  
Part 1: How are documents forged?

Images are everywhere—but which ones are genuine? At ICO-LUX GmbH, we focus on image-based insurance fraud. Whether purchased and digitally altered templates or AI-generated invoices, there are countless ways today to forge documents and, for example, fraudulently obtain money through private health insurance or supplementary insurance schemes. In Germany alone, the resulting losses amount to billions every year. And the cost of fraud? It is hidden within the price of the product.

It is time to launch a new article series on fraud and fraud detection. Over the coming weeks, we will explore developments across four key topics—each presented in ten concise theses.

Topic one:  
How do perpetrators forge documents?

1. Increasing automation and straight-through processing in insurance favour fraud attempts, as the majority of documents are now processed without human involvement.

2. Self-service portals and digital application processes lower the barrier to fraud—everything can be submitted online.

3. The perpetrator side is becoming more professional and better networked: guides and templates are systematically traded in fake shops, cost only a few dollars or Bitcoin, and—unlike what purely generative AI has produced so far—are deceptively realistic.

4. At the same time, image editing tools and AI image generators are accessible to everyone. With a strong focus on usability, they are becoming ever easier to use.  
What else?

5. In private health insurance in particular, offender profiling proves difficult. Well-established professional groups are also involved: teachers, civil servants, entrepreneurs, and even senior representatives of the Catholic Church have been caught. Losses often reach six figures, and fraudulent activity can continue over long periods before being detected without effective fraud detection software.

6. A particularly common modus operandi is the “adjustment” of invoice data, recipients, amounts, or entire documents through digital manipulation in PDFs, JPEGs, or PSD files.

7. The forged invoice is then printed out again and photographed. These photos are often deliberately of poor quality: tilted angles, shadows—posing a real challenge for analysis.

8. As fraud volumes increase and digital methods make fraud ever easier, swift action is required from insurers and banks.  
What does this mean overall?

9. The automation of application processes must go hand in hand with the automation of fraud checks.

10. By combining computer vision and state-of-the-art AI-based document forensics with a dedicated fraud department—as implemented by ICO-LUX GmbH—indicators can be detected in real time and fraudulent payouts prevented.

Image source: Mock-up by Freepik; fake shop screenshot and composition by ICO-LUX