The FIFRAUD project, which was launched in March 2021 to find more effective ways of detecting identity fraud through cross-referencing existing databases rather than relying on single use submitted customer data, has been declared an overwhelming success.
Co-funded by the Brittany region and in collaboration with the b<>com research institute, the project saw the implementation of groundbreaking anti-fraud functionality into our identity verification solutions after a research period of 16 months.
Thinking outside the data.
Traditionally, the identity verification process would only be carried out on information submitted by the user during the onboarding stage. The process did not include any cross-referencing with older data.
However, some types of fraud are not easily identifiable, and only become evident when compared to previously submitted identity documents, as fraudsters sometimes reuse identity elements from one attempt to the next. In isolation, these incidents can evade detection, and it is only when compared with previously used identity documents that they are flagged as fraudulent.
The aim of the FIFRAUD project was to find an efficient solution to this challenge.
Ambitious research integrated into production.
One of the main aims of the project was to minimize false rejections (documents wrongly considered fraudulent), while maximizing true rejections (documents rightly considered fraudulent). Given the newly introduced functionality, one of the main challenges of the FIFRAUD project was to ensure that the identity verification process with IDCheck.io was still able to be completed in under 10 seconds.
An important step in achieving the goals and overcoming the challenges was to design a machine learning algorithm that enabled fast and precise comparison between certain identity elements of the submitted document with the same elements of identity documents in the client’s database, after the client’s approval. IDnow’s biometrics research team was able to propose an innovative and extremely fast comparison method (1 millisecond per comparison) for the detection of fraudulent identity elements between new and existing documents.
In the course of the project, an in-depth analysis of the database over a 14-day period detected five repeat fraudsters, each of whom submitted multiple documents containing common elements. Thanks to the cross-referencing document analysis, the detection of fraudulent documents increased to above market standards, while the false rejection rate (i.e, documents that are considered fraudulent when they are not), was able to be kept well below the target rate.
Since the FIFRAUD research project concluded in late 2022, several of our customers have implemented the tried and tested improved anti-fraud functionality to detect fraud and enhance their identity verification process further.
Due to the success of the FIFRAUD project, research has been extended to add additional identity elements to the comparison step of the process.
Learn more about our highly configurable identity proofing platform.
Read more about ‘How to build a successful counter fraud culture in the UK’s private and public sector.’
Trend Report: Identity Fraud 2022.
Download the Trend Report: Identity Fraud 2022 to discover the latest fraud and identity theft trends, and how assurance verification solutions can enable organizations to protect the identity of their customers.
Get your free copy
