How Machine Learning can prevent identity theft

Online fraud was a hot topic of 2021 in the UK, Europe and worldwide, and it is going on strong entering 2022 as well, with countless identity protection services and synthetic fraud detection solutions on the market. The most common out of many types of financial fraud Ecommerce industry deals with is an identity fraud (real id theft or synthetic identity theft, when a new fake identity is created, instead of the existing one being stolen).

Synthetic fraud is usually characterised by the combination of different pieces of information belonging to different people, without their consent, that help create a fake identity (addresses, Social Security numbers, real names etc). Nearly half of the financial fraud losses of £1.26 billion in the UK, announced in April, 2021, is attributed to the stolen payment card information, which is a type of identity fraud.¹ Therefore, this type of fraud is a huge issue in the UK, and in this insight, we are looking at how it can be prevented thanks to machine learning implemented in synthetic fraud detection solutions. We discussed in depth the cybersecurity in Ecommerce in our whitepaper, where we dive into the impact fraud has on Ecommerce in numbers and how it can be prevented, as well as in the video dedicated to fraud prevention.

Difference between synthetic fraud, first-party and third-party fraud

What these three types of identity fraud have in common is the aspect of misrepresentation and the intention of deceit. First-party fraud (bust-out fraud) is performed by an individual that is being represented as themselves, however, their intent is fraudulent. For example, getting loans with no intentions of paying back, misrepresenting their income or status in order to get better loan interests, maxing out various credit cards from different issuers and then disappearing into thin air.

Third-party fraud (true identity fraud) is performed by a fraudster that misrepresents their identity and not simply their intentions, by stealing someone else’s id and using it in a complete form, without the real owner’s knowledge. Therefore, third-party fraud is a case of a pure impersonation.

Synthetic Identity Theft (Synthetic fraud or simply SIF) is the most complex one and can deceive many traditional fraud prevention technologies, since the false identity is composed of different pieces of real information that belong to different people and combined with fictitious data. This fraud is much harder to identify and track, since it is composed of numerous pieces of data, and therefore, is very damaging to the financial institutions, customers and businesses that lose fraud victims’ trust.

Synthetic identity fraud is using opportunities to navigate stealthily within the consumer-centered sectors, where traditional fraud prevention solutions are not capable of identifying the synthetic identity theft. A recent analysis by LexisNexis® found that 85% of synthetic identities were not flagged by traditional fraud models.² However, it can be fought against thanks to the advanced algorithms based on machine learning and the losses worth billions can be fought against successfully.

Machine learning and fraud trends 2022

The newest and most widely used development in identity theft protection is user-friendly machine learning (especially in financial institutions and Ecommerce). Machine learning algorithms are used in finance to detect and prevent fraudulent activities, as well as to automatise manual checks and processes, and the latter (process automation) is the most common use of machine learning nowadays. However, the advantage of machine learning implemented in identity protection services is that the AI becomes more sophisticated and smarter and performs better, the more it is used and the more exposure to data and fraud attacks it has, by learning through analysis of millions of data sets in a short period of time without prior complex programming.

Advanced AI and machine learning are widely and successfully used in biometrics. Machine learning manages to recognise fake images, such as customers’ selfies with an ID in hand, and as of late is used by many banks in the world as a verification of customer’s identity. This technology is able to recognise when the image is synthetic and if there are any discrepancies in the given information.

Another use of machine learning in finance and retail that is becoming trendy not only at the border controls in the airports, but also in the banks around the world, is NFC tags (Near Field Communication). In id protection, NFC helps authenticate the identity and if it is in possession of the rightful owner at the moment. The id document that has a chip inside, enabled by NFC, accompanied by a real time picture of the id owner is all that is needed for machine learning to prevent a real-time fraud from happening.

Machine learning in identity protection services and solutions helps identify easily patterns of a synthetic fraud that are very hard to notice by using traditional solutions and especially manually. This happens because machine learning acts not according to the preset rules in fraud detection, but according to the behaviours of a person that is being impersonated. It cross-checks the data available internally and online and predicts with high precision the outcomes of actions due to the behavioural history and millions of data sets collected over time thanks to deep and reinforcement learning.

Why it is important to keep up with fraud trends

During the worldwide pandemic in 2020-2021 the fraud grew immensely, due to the boom in Ecommerce and overall chaotic state in most industries, the fraudsters became more aggressive and started looking for more ways to steal funds from their victims. As a result, in the first 6 months of 2021, they stole £750 million from consumers and merchants in the UK alone, which is a 30% YoY increase compared to H1 2020, as stated by the UK Finance.³

However, fraud prevention solutions managed to save banks and their customers almost as much (£736 million). Nevertheless, identity fraud victims are three times more likely to leave their financial institutions or stop buying from the same websites only after one exposure to fraud, which destroys customer experience, trust and loyalty.⁴

This dynamic situation in evermore digitised world calls for immediate and effective solutions that would ensure the id protection and safety of the customers, and the more fraud develops, the more sophisticated solutions will need to be implemented on the basis of the fraud trends and the prediction of the direction it is taking. AI and machine learning uses are in constant development and advancement as well, therefore, the chances of protecting your customers through your company’s awareness are great, if you keep up with trends and with the way fraudsters think and identify blind spots, which also give opportunities to improve your customers’ experience and reduce friction. There are many solutions on the market that will keep you and your customers safe and will keep up with the trends for you, such as Payment Orchestration Platforms, TRA and others.

Subscribe to our newsletter to stay updated on the latest developments in payments and fraud.