During the COVID-19 pandemic, as an increased number of businesses have moved into the online space, the world has witnessed a sudden rise in fraud. To understand what types of fraud we are experiencing during the outbreak and what solutions are best to combat them, we spoke with Jimmy Fong, CCO at SEON, the fraud fighters.
PaymentExpert: Firstly, can you tell us a little more about SEON and your offering?
Jimmy Fong: It is clear from recent history that the fraud prevention industry has lost its focus and is quite publicly losing the fight against fraud. It’s in desperate need of more relevant data points to predict, not just analyse, behaviour, and also solutions that cater to the different needs of businesses.
This is exactly what SEON aims to achieve. Set up by digital natives straight out of the elite Corvinus University, the company was built out of necessity, after its two co-founders Tamás and Bence were inundated with fraud following the launch of their cryptocurrency exchange.
Despite the need for a solution that could be quickly implemented, they couldn’t find one. Instead, they found tools that took months to integrate, including lengthy contracts, with different solutions for different sized businesses – this simply wasn’t suitable for the exchange.
So, as two tech-minded entrepreneurs, the guys set out to solve the problem themselves, knowing they had to do things differently. As a result, SEON was born.
Now we offer a shift away from multiple fraud solutions from different businesses to one solution with multiple capabilities. Rather than analysing data at the front end of a transaction to identify instances of fraud, causing unwanted friction for customers in the process, our tools work in the back-end and analyse data from across the internet, to establish whether a customer or payment is suspicious.
These tools can be integrated into an existing business structure within minutes of hearing about the company – a stark difference from what is being offered by most providers. We also offer monthly rolling contracts, removing the need for fraud managers to make drastic decisions or multi-year commitments.
Payment Expert: How crucial can reducing the manual decision time be when it comes to halting fraud?
JF: We operate in a digital age and it has become increasingly difficult to make manual decisions as the human brain isn’t equipped to handle the vast amounts of data stemming from online payments. Instead, machine learning (ML) and artificial intelligence (AI) can sift through this data and reveal patterns that point towards fraud.
As a result, ML has become a core component of fraud management, as it is essential for operating in a bustling online space – meaning that reducing manual decision time is a must when it comes to combating fraud.
However, this doesn’t mean manual decision time should be removed altogether. This is because ML is only as effective as the rules that are inputted into a system and fraud trends are apt to change. Consequently, combining human oversight with ML will allow rules to be tweaked and enable companies to respond to fraud trends. Therefore, manual review should be increasing as a trend rather than decreasing.
Payment Expert: Can you tell us how the threat of fraud has changed as the world endured the global COVID pandemic?
JF: Taking the UK market as an example, research from Hiscox identified that an SME was successfully hacked every 19 seconds in the UK. But according to trends we have seen on our own platform, we believe this has risen by at least 50% in the past three months, amidst the pandemic. As a result, there may now be as many as 6,500 successful cybercrimes committed against SMEs every day in the UK alone.
The sudden increase in fraud during the crisis is first and foremost the result of a rise in online activity, as online fraud usually increases in correlation with it. This problem arose as customers have had no option but to shop from the safety of their own home during lockdown.
However, when it comes to assessing the types of fraud currently being experienced, there are several trends we are noticing.
Firstly, account takeovers (ATOs) have moved to the next level. At the same time fraud prevention teams and merchants aren’t taking them seriously enough, focusing their efforts on preventing chargebacks and transaction fraud instead of ATOS. This needs to change if businesses are to protect revenue.
Phishing is also now a growing threat and is responsible for 35% of major data breaches – that’s even before considering social engineering techniques, which also count as a form of phishing. Unfortunately, this type of fraud has been used throughout the outbreak to target those under financial pressure due to furloughs and job losses, with scams that offer financial aid.
The same people have also been targeted by false employment websites that steal identities. This year, both ID theft and synthetic ID fraud should be at the top of fraud prevention teams’ agendas, as they are being used against new services both in the US and UK.
What’s more, new security measures can inadvertently perpetuate data theft as they increase customer confusion. An example of this is the impact of new rules from the UK Gambling Commission, which force users to provide ID scans upfront. While born from good intentions, these measures create a huge demand for stolen and synthetic IDs.
Payment Expert: Are you able to detail further how you utilise AI when detecting and eradicating fraud?
JF: At SEON, we’re also pioneering the move away traditional AI and ML when it comes to fraud prevention.
While using such tools allows for greater efficiency and speed during the fraud prevention process, merchants need to be able to see why certain decisions have been made by these solutions, such as rejecting a customer’s payment. Current AI and ML-based fraud prevention tools don’t allow for this level of transparency and as a result, merchants are missing out on valuable insights about where their fraud is coming from. Consequently, they’re lacking the information needed to make effective decisions related to fraud.
Instead, businesses should utilise solutions that take a supervised learning approach. These tools provide valuable, actionable feedback about why decisions have been made and businesses can respond to this to ensure that genuine customers are accepted and fraud is prevented, all while maintaining high levels of efficiency.
Payment Expert: How has the pandemic accelerated the growth of AI within fraud detection?
JF: The global pandemic has really highlighted the need for AI, ML and supervised learning. With people flocking to the internet to shop and online transactions going through the roof, any fraud team that didn’t already have such solutions in place will be stretched beyond capacity, unable to process the millions of disparate data points and analyse them.
As a result, these teams have been forced to use AI, ML and supervised learning and we have seen adoption rates of these solutions accelerate throughout the pandemic. This will likely continue, as customers have become comfortable with shopping online during lockdown and will carry on purchasing goods this way as we head into the future.
