In my last post, I talked about how AI could be exploited by bad actors to enable chargeback misuse.
It’s true that large language models could be exploited to enable fraud and other illicit activity. But, that same technology could also be used to fight back against fraud, too.
In the wake of an alarming surge in payment fraud — losses came to $48 billion USD in 2023 alone — there has been concerted effort among payment companies to push back by leveraging new and existing technologies. We need to look more closely at AI as a potential force to help ward off threats and prevent invalid transactions.
Again, going back to my last post, we can take friendly fraud as an excellent example here, too.
The Current Problem
Let’s say a credit card transaction is disputed by a customer. The card issuer has to decide whether to issue a refund — otherwise known as a chargeback — to the cardholder. The dispute process was intended as a safety net for consumers in instances of fraud or unjust merchant practices. It’s crucial for consumer protection. However, misuse of this safety mechanism has had an adverse impact on businesses.
Friendly fraud happens when a cardholder contests a purchase without good reason to do so. This could be due to a misunderstanding, or because the buyer has the intent to commit fraud.
Friendly fraud can cause irreparable financial and reputational harm to retailers. These attacks cost them roughly three times the amount of the reversed transaction on average.
It’s critical that we, as an industry, secure consumers. But, as more consumers prioritize convenience, they tend to rely on banks to resolve disputes via chargebacks, which are escalating at an alarming rate. Chargebacks can hurt merchants’ relationship with customers and payment processors, leading to increased processing fees and endangering the survival of the company itself.
What’s Been Done So Far?
Friendly fraud can manifest in various forms, and to fight each instance can be a tedious, labor-intensive task without automated processes. For instance, say a customer alleges that they never received a product. The merchant can fight back through representment. However, this means they have to collate and present transaction data such as proof-of-delivery photos or recipient signatures from the delivery company, or authentication measures from their own systems.
Visa estimates that about 75% of all chargebacks are potential cases of fraud. This underscores the need for more efforts in distinguishing legitimate claims from those that constitute friendly fraud. Traditional chargeback management processes require employees to manually wade through vast amounts of data, costing businesses time and money while heightening the risk of human error.
The major card networks have reacted by introducing tools to help merchants cut down the number of chargebacks without infringing on consumers’ rights.
For instance, Order Insight aims to stop chargebacks before they are issued, allowing merchants the option of offering a customer a refund rather than going through the chargeback dispute process. At the back-end, Visa’s Compelling Evidence system can establish guidelines designed to streamline the evidence requirements for disputes. These are positive steps… but are not nearly enough on their own.
The Role of AI in Stopping First-Party Misuse
The automated aggregation of transaction data offers a solution to this problem.
This technology continues to evolve along with the trends in first-party fraud. AI systems can now detect fraudulent behavior patterns, allowing merchants to flag suspicious purchases and take action before the transaction is even completed. By using AI in their chargeback management processes, businesses can significantly reduce the risk of friendly fraud, resulting in fewer chargebacks and ultimately saving time and money.
AI capabilities have been a hidden advantage for years. The technology compiles key transaction data from various sources. AI can also suggest business improvements to address any shortcomings in procedure or policy. In doing so, it helps merchants be proactive about friendly fraud; something that was once thought impossible, given that friendly fraud is a post-transaction threat.
Furthermore, AI technology can also assist with identifying legitimate disputes from fraudulent ones. By analyzing vast amounts of data and detecting patterns or anomalies that may indicate fraud, AI can help businesses make more informed decisions when it comes to accepting or denying a customer’s claim for a chargeback. This helps protect businesses from potential losses, and also allows them to offer better service to their customers by quickly resolving any issues they may have.
The Future Is Bright
I’m very optimistic about the future of AI technology in fraud management. While first-party fraud continues to be a major issue for merchants, AI technology offers hope in combating this problem.
By using automated aggregation of data and advanced analysis techniques, AI can help identify and prevent friendly fraud, as well as assist with identifying legitimate disputes from fraudulent ones. And, as the technology continues to evolve, the role of AI in fraud management will only continue to grow, providing businesses with an invaluable tool to protect their bottom line and provide better service to their customers.
