Financial Crime: Banks betting on AI to win back trust

The Australian Financial Review  –  Monday 9 April 2018

International tech behemoth IBM is expecting a significant boost to its Watson Financial Services division in Australia, with the big four banks turning to artificial intelligence in its IT systems to win back the trust of consumers in the wake of the Hayne royal commission and years of banking scandals.

Speaking to The Australian Financial Review from his Astor Place office in New York, IBM vice-president of Watson Financial Services Marc Andrews said it was already conducting trials with most of the major banks in the country using its cognitive computer Watson for everything from reducing the strain of regulatory compliance, to overhauling the way the banks detect financial crimes.

“The royal commission will do two things – it will create heightened interest in being able to demonstrate back to the royal commission that the banks are doing something about these issues and that they’re looking to address them, and it might also reset some of their priorities,” Mr Andrews said.  “The banks have to be responsive and make sure they’re addressing any known issues.”

It is believed that NAB is running trials of Watson in the regulatory compliance space, while Westpac confirmed it was testing artificial intelligence technologies to support bankers in its consumer bank and the Commonwealth Bank head of emerging technology, Dilan Rajasingham, said AI was helping banks identify trends in data faster than ever before.

“Any emerging technology goes through a process of prototyping and validation,” Mr Rajasingham said.

“We’re focused on the simplification of processes and providing greater and more proactive insight.  We’re working with an array of solutions from start-up technology to quantum computing.”

IBM’s focus on developing AI solutions for banks began about 2½ years ago and led to its 2016 acquisition of Washington-based global advisory firm Promontory Financial Group, which had employed a team of former top banking regulators, including the first chairman of the Australian Prudential Regulation Authority, Jeffrey Carmichael, with the idea of blending the company’s cognitive computing abilities with the regulatory knowledge of experts such as Mr Carmichael.

Since then IBM has been piloting solutions with banks in three key areas – regulatory compliance, financial crimes, and anti-money laundering and “know your customer”.

Mr Andrews said it was estimated that by 2020 there would be 300 million pages of regulations for banks to comply with and Watson could help simplify this process.

To help address financial crimes such as fraud, manipulation of the bank bill swap rate, or insider trading, IBM is also developing a conduct surveillance solution that combines the traditional approaches of analysing trade patterns or communications into one holistic solution.

Mr Andrews said it could also be trained to detect and alert banks to poor cultures in teams, and help prevent bullying, drug-taking or sexism in the workplace.

Westpac consumer bank chief executive George Frazis said it was running a pilot program called Astro utilising IBM Watson as a virtual assistant to help bankers with “on-demand expert assistance in relation to home loan lending”.  “Astro ensures our bankers are equipped with the right knowledge to provide the most value for our customers in an environment with continuous regulatory and compliance changes,” he said. “While it’s early days, Astro will be able to learn and grow based on the interactions it has.”

Watson is also being trialled in the anti-money laundering space to speed up and enhance the due diligence process on higher risk customers, as well as for transaction monitoring.

Mr Andrews estimated that for every 100 alerts generated by most banks, less than 2 per cent were actually suspicious, but Watson had been proven to be able to identify 30-50 per cent of these alerts as false positives.

Watson had also been used to reduce the time it took for a banker to perform customer due diligence on a standard business from 13.5 minutes to 5 minutes and 20 seconds.

“We think there’s an opportunity to apply these techniques to provide banks with an alternative way to generate alerts that would be more productive and not generate false positives in the first place,” Mr Andrews said.

“We’re also seeing opportunities to collaborate and work with other banks to bring data together across banks to do things more effectively, whether it’s due diligence or managing alerts … an open banking scheme will help facilitate this and some of the initiatives from AUSTRAC are leaning towards this.”


The reporter was sponsored to travel to New York by IBM

Find out more about Indue’s financial crimes solutions.

Share this article:<br>Share on LinkedIn
Tweet about this on Twitter