The Good, the Bad, and the Ugly of AI in Payments

Remember when ChatGPT first came out and after the initial shock passed, you thought of the million and one ways this could actually help you and your business get stuff done? Buy-Now, Pay-Later (BNPL) giant Klarna’s co-founder and CEO Sebastian Siemiatkowski had the same idea.

Over the last month, Klarna’s OpenAI-powered customer service assistant had over 2.3 million conversations, doing the equivalent work of 700 full-time agents. It resolved customer queries in less than 2 minutes on average, compared to 11 minutes for human-to-human interactions, and works 24/7 across 23 markets speaking in more than 35 languages.

As Sebastian puts it, it’s a win-win-win for customers, employees, and investors, estimating the new AI bot will drive $40m in profit improvements this year.

“This AI breakthrough in customer interaction means superior experiences for our customers at better prices, more interesting challenges for our employees, and better returns for our investors.” - Sebastian Siemiatkowski


We well and truly live in an AI-powered world now, and this article explores the good, bad, and truly ugly ways we humans are utilizing it in the payments space.

AI’s very good at catching the bad guys

When it comes to fighting financial crime, this has until now been a situation of so much payment data and so little time. Today, AI-powered Machine Learning advanced pattern recognition algorithms are being used by card issuers, banks, payment processors, and other financial institutions to play the “spot the fraud” game. And it’s winning, with incredible speed and accuracy.

This includes screening payments in real-time and comparing to previously-recognized fraud patterns, like cashback fraud for example. Banks are also using AI systems to detect other behavioral anomalies indicative of scams, giving them “360-degree visibility of the behaviors of accounts on both sides of the transaction,” as ACI Worldwide’s latest Scamscope Fraud Report puts it.

But, as we know, AI’s not a one-trick pony.

Businesses are using the technology across the board, with J.P. Morgan this week reporting its AI-aided Cashflow Model has seen tremendous interest from clients after results found it cuts manual work by 90%.

More and more consumers are getting in on the action too, with AI-driven apps powered by Open Banking APIs helping them to better analyze their spending habits and manage their financials more responsibly. Upcoming PSD3 regulations will boost this consumer app ecosystem too.

So, safer payments, better fraud detection, quicker customer service, and greater financial management tools. What’s the bad news?

As amazing as AI is in payments, it’s not all sunshine and rainbows. The rise of Authorized Push Payment (APP) fraud pre-ChatGPT has only gotten worse, particularly among millennials and Gen Z demographics who use peer-to-peer (P2P) payments comparatively more.

"I'll pay for dinner and you just send me your half, right?" - heard at the end of every Gen Z restaurant experience.

APP scams happen when fraudsters trick someone into sending a payment to an account that’s outside of their control, like thinking you’re sending a payment to a friend or relative but instead to a fraudulent account elsewhere.

The UK’s Payment Systems Regulator said APP fraud accounts for 40% of all losses in 2022, and has recently announced consumer protection reimbursement requirements to ensure victims are not left out-of-pocket.

Unfortunately, ugly scammers are increasingly utilizing the tools of AI, particularly those of large language model (LLM) systems, to mass-generate convincing messages and phishing emails to vulnerable consumers and businesses.

Malicious actors have had to resort to such methods, as accessing encrypted and tokenized consumer payment information is increasingly, thankfully, becoming an impossible task.

Key Takeaways for Merchants

So there it is. AI is today already protecting your business by using advanced pattern recognition methods to save against frauds, while the proliferation of AI-powered customer service chatbots will soon also become a reality (which have equal levels of customer satisfaction to human CSMs too by the way).

But it’s not all good news. The rise of AI-powered scams and fraud, particularly in perpetuating APP fraud, is a growing concern. Fraudsters are using the same AI tools that businesses are using to protect themselves to carry out more sophisticated attacks.

It’s a constant game of cat and mouse, with businesses and regulators having to always find better ways to protect consumers.

As a merchant, it’s important to implement strong security measures like tokenization and encryption that can protect your customer’s sensitive payment information from falling into the wrong hands. Get in touch with Praxis Tech, a Level 1 PCI DSS-compliant payment orchestration platform to learn more on how to protect your business from malicious actors.