Australian Finance Regs Look To AI For Compliance

Australian financial regulators are feeling pressure to ensure financial institutions treat customers properly and have turned to artificial intelligence (AI) in an attempt to improve compliance, according to a report by the Financial Times.

The Australian Securities and Investments Commission (ASIC) spent a year investigating misconduct in the sector, and now it’s funding studies about how special forms of AI, including natural language processing technology, can help with sniffing out potential misconduct and improving oversight. 

The trend is starting to appear nationally. Spending on AI tech went up by $25 billion in 2018, an increase of 45 percent from the the year before, FT reported.

ASIC also wants to potentially use AI technology to scan online promotions, advertisements and financial planning documentation to look for rule-breaking and bad advice. Software would monitor conversations between customers and insurance sales workers to look for “hard sell” tactics or a failure to stick to disclosure rules.

“It could be the words used by sales agents or the way particular sentences are constructed that are problematic,” said John Price, ASIC commissioner. “This is not about automating the entire supervision process, rather it’s about risk targeting — identifying areas of concern that can then be handed on to experienced investigators for further examination.”

The ASIC inquiry found misconduct in the financial industry as well as a lack of enforcement. The report found some institutions charged fees for nothing, and sometimes to dead customers. They also lied to regulators and adversely cost customers millions by giving bad advice.

Many regulators around the world are already starting to use algorithms to get work done that usually fell on the shoulders of actual people. The National Health Service in Britain teamed up with Google AI company DeepMind to help look for breast cancer in mammograms.

The new technology could help in Australia by using a technology called “phrase-matching,” which would look for certain terms and patterns. Richard Kimber, founder and chief executive of Australian AI startup Daisee, is testing the tech by monitoring calls at Westpac, a large bank, and also at a life insurance company.

“In the past you could sample maybe 2 percent of calls, whereas now the AI technology enables you to sample 100 percent of calls,” he said. “This was a laborious and painful process for humans and it used to be too expensive to review all calls in an organisation.”