(This column originally appeared in Forbes)
We hear a lot about AI and what it will be doing in the years to come to (hopefully) better our lives and our businesses. But for some industries – particularly those in financial services – AI is real and already in use.
Your financial data is – not surprisingly – one of the top targets for hackers. While global cyberattacks have increased by more than 8 percent over the past year, the financial industry has seen a “staggering” 1318 percent increase in ransomware attacks with over half of financial institutions falling victim to at least one ransomware attack in 2021 alone, according to security firm SOCRadar. Other research has found that fraud at credit unions has also increased by a whopping 70 percent over the past year.
This is a battle not only for your financial security but for the viability of our financial services industry.
One of the biggest reasons behind this surge is the growth of AI development tools that are enabling attackers to increase and better target their attacks. But – to protect your data (and their profits) – the banking and financial services industry is fighting back, and with their own AI-based weapons. This is reality now for this industry…and the future for just about every other industry. So how are they fighting back?
For starters, many big financial services firms are digging deep into their own and others’ large language data models by training AI to learn about customers’ behaviors and ask questions like “is this something this customer normally does?” or “does this transaction make sense given the customer’s recent behavior?” or “is it reasonable that a customer just bought a product in London on the same day they were having dinner in Los Angeles?”
Credit card processing giant Mastercard, for example, is using AI to track the movement of scammed funds through “mule” accounts and analyzing account names, payment values, payer and payee history to identify unusual trends and fraudulent activity and then reporting its findings (and warnings) to member banks.
PayPal is rolling out software to “track all permutations” of addresses and personal information to predict consumer usage and identify unusual patterns – like patterns where criminals test out a credit card before using it – while alerting processors, banks and consumers.
Danske Bank in Denmark and JP Morgan Chase are both implementing deep learning AI tools to detect fraud, automate decisions (like freezing an account) and alerting security analysts. AI is also being used at these institutions to detect and quarantine malware and provide an early warning system for suspected attacks.
Validating External Data
As I’ve written in the past, data accuracy is critical when deploying AI. So even with all the automatic routines designed to spot trends, AI is expanding its language models to verify and validate internal data against external data sources before making its decisions. Because maybe a transaction that seems irregular isn’t so irregular when other sources are looked at, like sites that contain travel, buying and online behavior. I know this sounds creepy and there are all sorts of privacy concerns. But the benefit is added security and a better experience for customers because who wants their credit card declined when it doesn’t need to be?
Using Generative AI
Generative AI is also being used to defend against cyber-attacks. Many IT departments at banks around the world are using tools like ChatGPT to quickly write scripts and algorithms to help them neutralize incursions. Customized generative AI tools are also being used to create response plans, send alerts and provide – and even perform – actions when an incident occurs and even automate tasks like launching applications or automatically writing scripts to do things like automatically disabling unused accounts that may contain sensitive information which could be exposed by hackers.
Step Up Communications
Finally, AI is being deployed to improve communications at financial institutions, because the quicker everyone knows about a problem the faster it can be addressed. So when problems arise, new AI tools are learning not only what actions to take but what humans should be alerted. Reports and analysis are also able to be generated faster and sent to targeted individuals based on the data included. And of course – and assuming the data is reliable – AI communication tools are alerting customers and external parties when there’s a problem, requesting specific information to allow access, asking for responses via chat and email and helping to instruct customers to better protect their financial data.
Of course, there are many other uses for AI that don’t have anything to do with security – like enhancing customer service and automating wealth management activities – and that’s a topic I’ll take up in the future. But from spotting data trends to automatically performing tasks, the banking industry is leading the way in AI deployment because, when it comes to securing data, they have to.
Why? Because would you leave your money with a financial institution that didn’t do these things? Without this technology, our financial information (and money) would be at a much higher level of risk. As this industry breaks new ground in the use of AI it won’t be long before businesses in industries from manufacturing to construction will be using these same tools for similar uses.