Artificial Intelligence and Big Data : how data and insights can inform data protection

Oct 27, 2016

Bank fraud is something that is rarely out of the headlines. Fraudulent losses are the result of many errors, including insider threats, lack of security in online banking and card fraud, and they continue to rise. Card fraud has increased 19% year on year,  according to The Nilson Report, accounting for losses of around $16.3 billion in 2015. France has seen an 8.9% increase in card fraud and the USA, which has the largest fraud/loss ratio, currently accounts for 47.3% of the world’s payment card fraud losses.

Le rôle de l’intelligence artificielle et du Big Data dans la prévention des fraudesThe threat to banking is at least in part due to the explosion of data. It is expected that by 2020 we will be creating more than 44 times the data we created in 2009 - and that fraud will have resulted in losses of $35,4 billion. The storage and transmission of so much offers opportunities for fraud and cybercrime.

But as well as being part of the problem, it can be part of the solution.

The Evolution of Fraud Management

One of the most annoying things that can happen to a bank customer is to try and make a purchase and have that purchase denied while a fraud check goes through. We put up with it to avoid fraud, but it’s embarrassing and intrusive. Big data and Artificial Intelligence offers us a more intelligent and informed approach to data management. Such vast quantities of data offer potential for powerful and actionable insight, providing predictions based on real-time analysis of this data.

Knowledge is power.

Ensuring that customer protection is paramount, whilst also preventing normal transactions from being interrupted is a fine balancing act for banks. The evolution in handling fraud management can be conducted in a more intelligent manner by the use of big data - or ‘dataprints’. Dataprints are like fingerprints - they give us unique information about a given person, action, place and point in time. Using these digital fingerprints to provide accurate identification and analyse it through Artificial Intelligence leads to better fraud detection. Analyzing the behaviors of transactions and devices and identifying when the usual patterns have been disrupted provides a warning sign for banks and customers.

Analysts McKinsey in their look at disruptive technologies, predict that neural networks will utilize big data to enable “knowledge work automation”. This can be applied in a much more accurate and pro-active way to detect fraud, looking ahead to prevent it rather than retroactively solve The use of these neural algorithms through artificial intelligence software means that the process can be improved through learning and applying new and more refined algorithms, ever improving its sophistication and capabilities.The recurring activity means that the machine’s ability to predict and classify information advances, making it easier to make data-driven decisions.

It’s all very well to say that data and technology can help prevent fraud - but what does this look like in practice, and how can banks achieve this?

1. Collection and Centralization of Internal Data

The trouble with big data can be found in its name - there is a lot of it. It is necessary to devise ways of collecting and storing this data in a manner that allows you to take full advantage of it when you need it - but also keep it secure.

Normally, data is created and held in silos. It is typically collected in a division / department / business area, type manner and because of this delocalization, it ends up being difficult to collate, distribute and utilize in any sort of global way. Centralizing the collection and management of data means that you can more easily access the data and cross-reference it, which creates the basis for a more intelligent and focused approach for the use of data. A July 2014 survey of bank respondents by The Economist, found that half had applied centralized analytics to big data management through artificial intelligence software. The banks that applied this centralized approach were seen as having the most holistic approach to risk mitigation and fraud prevention, and been able to enhance their security as a result. It is something the industry needs to move forward with as a whole in order to maintain its competitive advantage as well as fight fraud.

The reason why the centralization of data is important is because it is the first step in creating intelligent big data – data that works for a bank and the customer by allowing them to understand customer profiles and typical customer actions. This enables them to not only mitigate fraud, but service their audience better. The implementation of big data centralization is as much a process as a system and requires synthesis of legal and regulatory compliance, a security and privacy focus, strong management and the best technology.

2. Leveraging External Data

Big data means information from multiple and often highly disparate sources. One of the new challenges for data collection have arrived in the form social media platforms like Facebook and LinkedIn.

Analyst firm, McKinsey has shown that the use of  external data, such as social media activities, can have up to 35% improvement in areas such as risk mitigation, as well as allowing the development of better insights into customer behaviour and ultimately in fraud behaviour analysis. One of the reasons for lack of uptake in this area is the difficulty of retrieval of such data. Although this is certainly achievable in terms of technology through the use of social graph APIs. However the consent and release of this data is often a legal minefield and customer privacy worries and media scares themselves can be a hurdle to jump.

However, external data can be an extremely useful tool in the fight against fraud whilst still upholding privacy concerns. Using big data analytics and especially external data tracking, can allow a bank to spot fraudulent activity. For example, modelling of transaction patterns can be performed without the need to identify individuals. This results in a privacy enhanced approach and using external big data patterns, such as tracking the use of credit and debit cards, as well as IBAN accounts, allows banks to spot fraudulent activity quickly and in real time.

Going forward into an era of instant payments, external data tracking that is conducted in a privacy enhanced manner will become even more important. The ability to keep track of these payments, whilst ensuring personal data is obfuscated, all in real-time is a challenging but ultimately empowering new tool for the industry.

3. Using Behavioural Profiles to Prevent Fraud

Big data is revolutionizing the process of ‘Know Your Customer’ or KYC. As KYC becomes KYCd, or Know Your Customer’s data, a more accurate and in-depth approach to consumer understanding can be rewarded by more impactful anti-money laundering (AML) and other types of fraud detection.

Being able to model patterns of behaviour by using predictions based on internal, external and social big data is transforming banking.It not only gives you insight into normal behaviour, but that baseline then allows comparison and identication of patterns, similarities and differences - and fraud. Technologies such as geolocation, can be added to the arsenal, so those incidents when a customer is interrupted from making a legitimate purchase are greatly reduced, whilst real crime is detected.

Having deeper knowledge of your customers from behavioural analysis based on a widely collected set of data, will be beneficial in terms of customer insights and tailored responses, as well as fraud prevention. However, it can also offer challenges in terms of security and privacy. Customers are now more informed about privacy considerations and have become less happy about sharing their personal data with any company, not just a bank. Sopra Banking Software report found that 80% of customers would be willing to share their personal data, as long as they did so using a consented, ‘opt-in’, approach and in doing so they were incentivized by better rates and so on.

New EU privacy and data protection laws, which are an adaptation of the Data Protection EU Directive 95/46/EC, are due to be finalized this year. The new data privacy laws will be more restrictive and will have focus on, for example, data stored in the Cloud.This requires a Privacy by Design (PbD) approach when creating Cloud based systems, especially those that store, transmit and transact data. Handling these more extensive regulations needs a more rethink in the approach to security and privacy.

Big data gives us big solutions to the issue of fraud. It is a powerful tool in the fight against crime, including cybercrime, but it needs to be handled in a way that optimizes its use, secures the information and takes personal privacy into account. Although the collection of data, how to centralize and manage it, how to make it safe and how best to analyse and make predictions from it are all challenges, they also offer huge potential.

The digital revolution that has brought us big data can also bring us big banking.

David Andrieux, Senior Business Consultant

Related Events

No related event