AI is a powerful tool in fighting financial crime, with usage set to triple by 2021. But what about the risks?
AI has the potential to revolutionise every aspect of daily life. Businesses use AI for recruitment, customer support, sales and more. And it’s fast becoming a vital part of security surveillance – tackling financial crime like fraud and money laundering.
The National Crime Agency (NCA) believe hundreds of billions of pounds are laundered each year in the UK, so it’s a huge problem.
Not surprisingly, the big banks are in the best position to make use of AI to fight money laundering, according to Jim Gee, Head of Forensic Services at Crowe.
They have access to vast volumes of data, and AI is ideal for trawling through massive data sets looking for patterns and anomalies. Suspicious activity can often be flagged up in real-time.
“With technology used more effectively and creatively, you can reduce manual and repetitive tasks, and humans can focus on the more material risks,” he adds.
Limited, but effective, capabilities
However, Adam Williamson, head of professional standards at AAT, says AI’s ability to deal with big data is the “extent of its capabilities” right now.
It hasn’t reached a level where it can make complex decisions. But it’s excellent at data mining and helps accountants make connections and perform due diligence.
José Hernandez, forensic accountancy specialist, author of Broken Business and Founder and CEO of Ortus Strategies has worked on some of the largest fraud, bribery and money laundering cases on record.
He says it “wouldn’t have been possible” to gather relevant facts and evidence relating to the cases had it not been for AI.
“Each of the significant internal investigations we’ve been involved in have employed very sophisticated digital forensic tools. They used AI to organise, search and analyse vast quantities of data like emails, chats, text messages, calendar entries and financial records,” he says.
“These cases often involved a dark web of third-party intermediaries and offshore shell companies. Without such tools, we wouldn’t have been able to separate signal from noise and identify patterns quickly and efficiently.”
Finding hidden criminals
As Hernandez explains, AI tools play a crucial role in helping organisations identify hard-to-detect forms of criminal activity.
AI can also help identify modern slavery practices and human trafficking, says Williamson. It comes into play when gang leaders bringing trafficked people into the UK open up multiple bank accounts in numerous branches (often on the same day).
“It’s a common smurfing technique, where small amounts of money are put into multiple accounts as a way of hiding larger amounts of money,” Williamson explains.
“Someone might be paid £500 every week, or there might be regular, multiple payments to travel agencies and low-cost airlines. On their own, such transactions may not necessarily be suspicious, but together, they can point to trafficking and other unlawful activities.”
Complete your AML survey
If you are an AAT licensed member who is supervised by AAT for AML, your firm must respond to our AML survey by Monday 6 December. Click below to find out more:
Often however, AI’s ability to flag up potential criminal activity relies on information sharing, particularly as gangs will open up accounts across different banks. But Williamson says there’s a general “market reluctance” to share information, not least because of ethical and confidentiality issues.
“If you have limited data to work with, the system can’t make the connections and join the dots,” he says. “But banks aren’t always happy to give out customer information.”
However, the Fifth Money Laundering Directive – due to come into force in January 2020 – focuses on transparency and direct access to information. It will require banks to hold registers of bank accounts, their owners and beneficiaries, so AI systems will hopefully have more access to vital information.
Ironically, ethical issues are arising from using AI tackle this type of crime, and not just around GDPR and information ownership.
Williamson warns there have been countless examples of inbuilt or learned bias from AI because it hasn’t yet got a strong enough cognitive ability. Limited data sources and crude programming can result in discriminatory conclusions.
AI is learning to stereotype.
He uses the example of loan approvals, where people from a particular demographic or background may be turned down for loans despite meeting the criteria. It happens because the system has “learned” from previous process outcomes or is using limited data sources to determine risk. “AI will exacerbate any inbuilt or learned bias,” he says.
There’s also the issue of AI itself being used to actually commit crime.
It becomes a bit like financial chess, says Williamson, with two systems pitting against each other, learning to out-manoeuvre the other.
“There’s a continual learning process between both parties,” Jim Gee adds. “Those perpetuating financial crime will be using AI to increase their chances of success.”
The most significant impact on fraud and financial crime, he says, has come from changing the balance of human behaviour: mobilising and growing the “honest majority” and deterring and shrinking the “dishonest minority”.
Gee insists that the future of AI lies in identifying weaknesses of criminal systems – every flaw removed is another financial crime stopped. “Ultimately though,” he says, “AI needs to prevent crime, not just detect it.”
AI is streamlining fraud detection and getting ever more efficient and creative. With the Fifth Money Laundering Directive hitting the scenes in Jan 2020, the hope is that this results in much more access for AI systems. And with larger data sets to comb through, AI should exponentially improve with crime detection.
From here, the next step is developing further towards crime prevention.
Read more on financial crime and anti-money laundering here:
- Drug dealing, money laundering and crime – just another day in the office? (coming soon)
- Know your anti-money laundering responsibilities inside out
- When clients don’t tell you everything…
AAT Comment offers news and opinion on the world of business and finance from the Association of Accounting Technicians.