AI for compliance officers

When many people think of AI (artificial intelligence), they think of Terminator, a human-crushing, world-ending, indifferent machine. When almost anyone thinks of paperwork, they think soul-crushing, cumbersome, tedious work. What if AI, instead of destroying people, destroys paperwork, ending the drudgery of mindless work?

No, we’re not talking about some Hollywood sci-fi flick, but rather introducing AI into GRC (Governance, Risk and Compliance). Banks, and other financial institutions, have huge reporting requirements around Anti-Money Laundering (AML) and other legislation. As one unnamed executive states: “There are so many manual processes in AML … People are taking pictures of documents all day long and grabbing information from different systems. A robot could alleviate this drudgery.

We’re not talking about doing strategy, oversight, innovation, interpretation and other thought intensive processes; that is science fiction at this point. Rather, we’re talking about the many manual processes that take up the vast majority of a compliance officer’s time.  A 2014 Booz Allen Hamilton report found that: [tweet_box design="default" float="none" excerpt="#AML analysts typically spend 75% of their time in data collection and another 15% in data organization and entry"]AML analysts typically spend 75 percent of their time in data collection and another 15 percent in data organization and entry.[/tweet_box]

As opposed to eliminating employee positions, AI and other automation tools can assist employee productivity, allowing them to focus more on the high-level requirements. One example is using AI to parse long, complicated regulatory legal documents and distribute tasks appropriately. Consider the Dodd-Frank Wall Street Reform and Consumer Protection Act; that one piece of legislation is over 3,000 pages. On top of that there are amendments, guidance, rulings, and interpretations that occur on an ongoing basis.

All this is only one piece of legislation and there are many others, from different regulators and different jurisdictions. Considering that many financial entities operate internationally, the amount of rules go up exponentially. It’s much more productive to have an AI to keep on top of all these requirements; employees can spend their efforts actually assist in complying with all these rulings, as opposed to just trying to understand what they need to do

Another example is in the realm of money-laundering investigations. With so much time spent on data handling, that leaves much less time for investigations. That puts a lot of pressure to quickly go through cases and potentially overlook criminal activities. If AI could quickly sort through and flag more complex situations for manual review, investigators can then focus on these more serious cases.

There are many other scenarios where the power of AI to quickly analyze large data sets will assist compliance. Detecting unauthorized trading, flagging unusual accounts, and catching breaches of conduct are all areas where AI can help.

But, AI can only work on clear patterns and it’ll be up to actual people to figure out cases in the grey area. More subtle scenarios still require the power of human judgment and thought; these are more interesting anyways.

Know Your Customer

Consider the requirements to Know Your Customer (KYC). Clients using Trulioo’s ID verification service can enter criteria to decide if an ID check passes or fails. But that’s not the end of it; if someone fails, for example, a compliance officer can check why they failed. Maybe they changed their email address, so that doesn’t match and further investigation can determine the new address is legitimate. The automated system allows most checks to be done in seconds, only requiring compliance personnel to review the exceptions.

For example, nTrust teamed up with Trulioo to bring additional online authentication services into their fold. By utilizing Trulioo’s global AML Watchlist and profile verification services, nTrust is able to make decisions about how much risk it is willing to take on for a member before it has to deliver any friction.

If an individual is flagged for any reason, nTrust is then able to place the account under review, and nTrust can examine its own risk appetite and all the business rules set out to determine if it’s willing to take on the risk for that individual.

“It saves a lot of time, which is huge,” nTrust Fraud Management Specialist Darcy Berringer noted. “Before I was manually uploading and checking lists and subscribing to every sanction list you could possibly imagine to get updates.”

While not technically AI, the result is the same; compliance can focus on the more intellectually demanding tasks and don’t need to concern themselves with the boring, mind dulling aspects of handling data. Automation — AI, robots, smart agents, etc. — does promise a Hollywood sci-fi scenario, but it’s not a dystopian, destructive doomsday scenario but rather, one of purpose and productivity that frees us from the drudgery of rote, unimaginative work.