In the news:
Tryg: https://m.finanswatch.no/artic...
Frende: https://m.finanswatch.no/artic...
Fremtind: https://finanswatch.no/nyheter... | https://e24.no/shared/privatoe...

About the problem
Fraud is a persistent and common problem in the insurance industry to the cost of companies, customers and society. According to Finans Norge, detected fraud cases in Norway amounted to almost 500 million NOK in 2019 - and those are just the cases that we know about. The amount is likely much larger if you include the fraud cases that are never discovered by insurance companies.

To address this issue, NCE Finance Innovation has been working with the Norwegian insurance companies Tryg, Fremtind and Frende to create a more efficient safeguard against fraud based on collaboration and data-sharing. Along with partners Webstep and PwC, the group has spent the last months looking into an appropriate technical infrastructure and legal framework. The project participants have now agreed on a workable design to test in a pilot study and focus has therefore moved to building the solution and sharing the needed data to test its value.

The idea of the service is to share anonymized data on a closed cloud platform. Individual companies will not be able to see each other's data, but will be able to use the larger data set for training their own fraud-detection models through machine learning. The goal is to have access to more training data and thereby create models more efficient at identifying red flags and detecting potential fraud cases.

"At Tryg we have a highly skilled team of data scientists working with fraud and claims analytics. But modelling skills can only get you parts of the way. In order to improve, we are completely dependent on access to large volumes of high-quality data. Fraud is a difficult one. The insurance companies are all sitting on their own pile of data, painstakingly aware they can only observe parts of the true picture. By sharing anonymized data in a safe manner, we can all gain new insights and improve fraud detection. This will benefit our customers, the insurance companies and society in general." says Anders Daasvand Sleire, Senior Data Scientist at Tryg.

This sentiment of creating value for both customers and society is shared by Roy-André S. Lyngbø, Head of Counter Fraud & AML at Fremtind:

“Fremtind’s models for digital risk assessment are crucial in our combat against fraud and anti-money laundering. Through this project we hope to develop more accurate risk models which will help us provide excellent customer experiences and fight crime at the same time. In the end this joint effort will benefit both our customers and the Norwegian society.“

The initial pilot will be operational later in the fall. At that point the participants will evaluate the lessons learned to decide how to proceed with the next iteration. The ultimate ambition of the project is to have a wide impact on insurance fraud in the industry. For this to happen, more companies will need to join the current group.

- "Both industry players and government should have an interest in battling insurance fraud. We hope that, as the value of having larger amounts of training data becomes apparent, this solution will eventually become an industry standard - perhaps even mandatory for all insurance companies." says Henrik Hedegaard, Project Lead at NCE Finance Innovation.