Fraud a natural fit for AI applications

By Michael McCaw | 26 September 2019

Detection of fraudulent activity in the payments world is the perfect place for artificial intelligence to blossom, according to Riteesh Singh, senior vice president at Finastra and Joël Winteregg, CEO and co-founder of NetGuardians.

The pair were speaking on the side lines of Sibos in central London this week, after announcing a partnership between the two companies to launch an AI powered fraud detection solution for financial messaging.

“Any area that is data heavy – fraud prevention is the perfect example – provides great opportunity for AI,” said Singh. “Another area is anywhere that’s people heavy we’ll see more and more advances. Banks back offices are going to find some great applications.”

In 2018 in the UK alone, fraudulent activity against banks sat at £1.2bn, up from £967m the previous year, according to figures from UK Finance. A number of banks and larger fintechs are seeing the benefits of working with specialist AI firms, but that wasn’t always the case according to Winteregg.

“Banking and financial services is far more open, today,” he said. “They’re far happier to work with specialised vendors. The timing is right, banks really fear new players, with neo banks coming into play, and so they have to quickly find the right solution.”

For Singh the fact that financial services firms are reaching out to smaller players for assistance with specific challenges is logical.

“Fraud and regulatory reporting – banks are far more open to collaborate as they want to protect themselves," said Singh. “We want to bring the latest technology and the latest innovations to our clients. We partner with fintechs to do that.”

For Winteregg, AI’s benefits are developing fast.

“With AI in fraud you have plenty of companies working within cards and merchants with fraud challenges – it’s very common. It’s huge in fact. There are many companies looking at the backend of the payment – when your card payment is going through the money transfer later down the line. With AI you need to learn. Without learning there’s no value - it’s just analytics or reporting. With card fraud there’s a lot of data assets to learn from. There’s thousands of instances of fraud each day for card providers.” However, AI can be far more focused – and needs to be.

“But if you take the Bangladesh case, there was just one instant of fraud in ten years and it was huge. When you have a small volume of fraud you’ve got little to learn from. We’ve developed a way to learn with very few instances of fraud,” said Winteregg. “What we’ve developed is a way to monitor the journey in a way with very little amount of fraud.”

For both, the market is only just starting to consider the benefits and use cases of AI, with regulatory reporting and customer advice strong possibilities once hurdles can be overcome.

“The systems are getting more and more advanced," said Singh. “For things like regulatory reporting liability has to be fixed and that’s the thing that regulators need to consider.”

Become a bobsguide member to access the following

1. Unrestricted access to bobsguide
2. Send a proposal request
3. Insights delivered daily to your inbox
4. Career development