Whether it’s renting a car, making purchases, making transfers, concluding insurance contracts or opening an account at a bank: today, business transactions are increasingly being carried out online. The background is not just the widespread use of smartphones and tablets. There are also increased demands on the customer journey in terms of speed and convenience.
Companies have to adapt to these customer needs. At the same time, however, it must be ensured that the high performance does not result in any security gaps. Artificial intelligence (AI) can help meet compliance requirements without affecting process speed. Three fields of application where the use of AI can close potential security gaps:
Numerous companies such as car-sharing providers, dealers or companies from the gaming sector are legally obliged to carry out a customer check. This applies to traditional business as well as to online contracts. Car rental companies and gaming platforms must ensure, for example, that the contractual partner is identified and has reached the legally required minimum age. The identity check is also relevant for operators of online shops. It is essential to prevent fraud cases such as delivery to bogus addresses and non-payment – especially when buying high-priced items.
Relying purely on the information provided by a user at this point is, of course, risky. A review of the data by employees sometimes takes several hours and is therefore ruled out from the point of view of the positive user experience. Automatic processes based on artificial intelligence make it possible to carry out identity and age checks within a few seconds. Modern solutions are supplemented by biometric systems and can make verification even more reliable.
The advantages of this approach are apparent: Customers save long waiting times and can complete their transaction with little effort regardless of location and time. There may even be price advantages due to the omission of the provider’s risk premium. Companies, on the other hand, realize enormous cost savings and minimize compliance risks without endangering conversion.
KYC Onboarding Processes At Banks
In the financial services sector, too, customer-centricity is the order of the day. One of the essential tasks in this context is significantly shortening the KYC process as part of the onboarding. After all, the manual collection, processing and interpretation of customer data sometimes take several hours or even days. Despite this process complexity, the quality and completeness of the processed data are often not guaranteed without exception. This approach is neither extremely safe nor efficient. It also does not do justice to the customer’s expectation of a high response speed.
At this point, AI-based identification processes come into play again. In the case of banks, in particular, corresponding solutions can be added upstream of the video identification process to identify and filter out security anomalies in advance. The results are checked and transmitted within a few moments. For financial service providers, this procedure represents an additional security component.
At the same time, cost savings result from eliminating manual steps and avoiding unnecessary identification processes. On the customer side, the approach offers the advantage that the data provided is verified before the video session, ensuring a smooth process. In addition, the entire process is accelerated.
There are already solutions that have proven themselves several times in practice for the first two scenarios (identity check and KYC onboarding processes). An AI application area that is still at the beginning of technological development but still has enormous potential is fraud detection. Identity verification is fundamentally relevant for all industries but is particularly explosive in the financial services sector. Incidents such as credit card and insurance fraud represent an exposed risk.
Cases of fraud are detected through the detection of anomalies. Artificial intelligence is predestined for this task. Based on training data, she learns the “Normal state” of a particular event. If processes then deviate from the typical behaviour, the AI sounds an alarm. The success of this approach depends in particular on the underlying database. For example, to detect credit card fraud, it is necessary to provide the AI system with the transaction history of one or more banks in advance.
Previous claims reports, correspondence and decisions or metadata from social networks, and localization data from mobile phones could be used to identify insurance fraud. In the classic scenario of fraud detection, if-then rules usually apply. A limitation of this type could, for example, be as follows: If the same credit card is used to withdraw funds at five different ATMs within two hours, then there is possible fraud. As a result, the card would then be automatically blocked.
However, if the fraudster changes his behaviour, then if-then rules are useless. On the other hand, Artificial intelligence does not need any rules but recognizes (new) patterns and deviations in the data independently. In the race between security measures and new fraudulent schemes, it represents a considerable time advantage. In addition, AI delivers significantly fewer false-positive results than the rule-based approach.
AI-based solutions for fraud detection are, of course, not limited to the end customer business but are also suitable for combating organized financial crime. Corresponding systems are able, among other things, to identify patterns of suspicious cash flows. This, in turn, is an essential measure against money laundering.