We’ve all heard that old adage about the decades where nothing happens and the weeks where decades happen. I’ve been thinking about it a lot recently, and how it aligns so perfectly with the field of technology. If the last year has shown us anything, it’s that this sector has the potential to enter periods of intense innovation in the blink of an eye. A great example is the dramatic, recent rise of AI and cybercrime.
The Advent of AI
Across a myriad of sectors, AI is on the rise and is bringing new solutions to the fore, which are actively shaping the world of tomorrow, today. Companies like OpenAI are central to this drive and have provided a framework that’s allowed mainstream AI adoption to take place for the first time. From the outside looking in, it’s certainly an exciting time to be involved with the sector, but it’s a trend that also comes with some drawbacks.
You see, such has been the speed of adoption of AI systems in recent times, that many people are unable to recognize the damaging ways that these solutions could also be leveraged by nefarious actors. Sadly, all the early indicators show that AI tools are going to be just as useful for crooks as anybody else. We’ve already seen evidence to support this worrying growth in the world of online fraud and financial crime.
Tackling the Trends
Fraud is on the rise across the internet, with a noticeable spike apparent in the past twelve months. Clearly, growing economic pressures are pushing more people toward this arena of online crime, including online fraud and money laundering.
However, perhaps most worrying is that fraudsters are now targeting larger amounts of money with their online attempts and are doing so in an increasingly sophisticated manner. In this pursuit, online criminals are now leveraging more advanced fraud tools, such as AI technologies and large language model solutions, such as ChatGPT-3, which are making their attacks more efficient and effective.
An Unwelcome Evolution
Whether it’s using machine learning in combination with fraudulent bots and other automation methods to create more efficient attacks that better evade rudimentary velocity checks, or utilising large language models to create bespoke, differentiated phishing attacks that are infinitely more believable, fraud is evolving online. For the first time, we’re seeing how powerful these technologies can be when they fall into the wrong hands.
I’d go as far as to say this evolution has led to the rise of a new type of fraudster, which fraud expert Frank McKenna has previously referred to as ‘shapeshifters.’ Using AI and other tools, fraudsters fitting this description constantly shift and shape their approach to fraud to better exploit gaps in defenses. What’s more, these technologies allow fraudsters to differentiate attacks to avoid detection and improve the effectiveness of their attempts.
Put simply, this shift constitutes a lowering of the barriers around fraud in a moment of acute economic concern. When combined, these two issues have created an explosion in online fraud, which very few companies had prepared for. For some context, between September to December 2022, we found that the average amount of money that fraudsters were trying to steal per transaction had increased by 300%.
Similarly, between October to December 2022, we also saw a 12.9% rise in suspicious user activity across our entire platform. Additionally, our research also indicates that the average fraudster is now using more sophisticated types of tech to hide themselves online. Ultimately, there is little doubt that fraud is evolving online and that those committing these acts are becoming more brazen, refined, and successful.
Fighting Fire with Fire
The good news is that modern fraud prevention solutions can mitigate this threat before it grows any further. The key is to prioritize platforms that utilize AI technologies, including whitebox and blackbox machine learning modules, to tackle fraud. Right now, these are the only solutions with the power needed to disrupt these more advanced fraud campaigns and to stop online crooks in their tracks.
We’re already seeing the effectiveness of this approach across our own platform. Since September 2022, the use of whitebox machine learning rule suggestions has increased by over 30%. In the same time period, SEON’s blackbox machine learning module – which provides a separate fraud score to inform decision-making – is being used 46% more. Thankfully, by taking this approach and choosing to ‘fight fire with fire,’ companies can greatly reduce the fraud risk to their business and their customers.