Both government and market regulators are increasingly using artificial intelligence (AI) and data analytics tools and techniques as part of their oversight functions, including surveillance and auditing. These watchdog groups say they are looking at building out their AI initiatives in the years ahead.
“We’re actually piloting using AI to calibrate our alerts and reduce false positives, so that our surveillance officers are focused on the more substantive alerts,” says Victoria Pinnington, senior vice president of market regulation with the Investment Industry Regulatory Organization of Canada (IIROC) in Toronto. The regulator has a dedicated data analytics team that has been streamlining data processes and looking for ways IIROC can apply AI technology.
Over the past few years, the Canada Revenue Agency has been using data analytics and AI, such as machine-learning algorithms that predict tax non-compliance and detect activity in the underground economy. Since 2018, the Department of Justice Canada has licensed the use of Tax Foresight, AI software developed by Blue J Legal Inc. in Toronto, which employs machine learning to predict – with about 90% accuracy, according to the company – how a court might rule on a particular tax scenario.
“It’s not just about speeding up [analysis] that would otherwise happen,” says Benjamin Alarie, co-founder and CEO of Blue J Legal and Osler Chair of Business Law at the University of Toronto. “It’s about making [widely] available a really good prediction that would otherwise be the domain of an experienced [lawyer].”
AI technology could bring more certainty to the interpretation of tax law, Alarie adds: “Everyone benefits from that.”
AI allows computers to replicate some features of human reasoning, including analyzing data, identifying patterns and adjusting for new information. In general, government and institutional regulators still are in the early stages of introducing AI and related machine-learning initiatives. Many government and institutional regulators are either piloting AI projects or have indicated they are considering developing AI capabilities, not only to become more efficient in how they perform their roles, but also to keep pace with the market participants they oversee.
“We know our dealers are starting to invest in this area; we know their clients are investing in this area. So, we need to be responsive,” Pinnington says.
In May 2019, IIROC launched an upgraded market surveillance system that uses Nasdaq’s SMARTS. “[This upgrade] will mean that we’re being far more proactive and efficient in detecting bad behaviour in the market, and that cannot help but enhance confidence for investors and advisors,” says Pinnington.
As for the Mutual Fund Dealers Association of Canada (MFDA), the organization states it has leveraged technology and data to identify patterns and target issues to enhance suitability testing in examinations. The MFDA declined to comment on whether it’s considering AI initiatives.
Over the past few years, provincial securities commissions have been looking to technology for help in their oversight roles. These efforts include launching regulatory development “sandboxes” to help, and learn from, fintech firms, and working on a new market analytics platform to allow for more efficient data analysis. Provincial securities commissions are also co-operating with international regulators to foster regulatory technology innovations.
The Autorité des marchés financiers (AMF), Quebec’s securities regulator, has been developing AI technology prototypes in its in-house fintech lab. The lab developed CP en temps réel!, which uses natural language processing to review companies’ news releases in real time. The technology ranks releases by probability of having misleading information, says Lise Estelle Brault, senior director of fintech, innovation and derivatives training with the AMF.
“Rather than reviewing 10 press releases randomly, [AMF] analysts now can review the 10 that the machine has decided as being the most problematic,” Brault says. “So, we’ve reduced the number of false positives.”
Regulation experts say AI could allow regulators to be more proactive in alerting market participants to issues before they become a problem.
“A high percentage of enforcement cases are about unsuitable investments,” says Colette Arcidiacono, founder of Montreal-based compliance consulting firm Conformité 101, and formerly case assessment manager for complaints received by IIROC. So-called “regtech,” including AI, would help “at the beginning, when the portfolio starts not to fit with the client’s objectives.”
Yet, experts don’t anticipate these innovations will replace human analysts and financial advisors, whether in the regulatory realm or in the broader financial services industry.
“It’s not about substituting one form of intelligence for another,” says Rhodri Preece, senior head of industry research for CFA Institute in London, U.K. “It’s about building greater collective intelligence by harvesting the cognitive powers of machines and humans. “