AI in transaction monitoring – Two birds, one stone
Regulators are often typified as conservative, sometimes constraining organisations that exist to put limits on the extent to which banks can innovate and trial new solutions. However, regulators worldwide do seem to be rapidly changing their approach, and banks can act swiftly to take advantage of the new outlook by using regulatory change to stay ahead of the competition.
There have been several indicators in recent months that regulators are beginning to encourage the use of innovative technology amongst financial institutions. For example, the FCA are currently hiring for a Director of Innovation tasked with driving ‘engagement with technological innovation within the financial services industry’. The UK regulator also provides a sandbox environment that allows businesses to test innovative propositions. Banks can use waivers and modifications to existing rules in order to quickly understand the positive benefit that technology can have on customers, whilst also getting immediate feedback on its regulatory suitability.
This open, collaborative approach is hugely beneficial within the industry as there are many areas within Financial Services which are heavily regulated but have the potential to make real use of innovative technology. One such area is Transaction Monitoring within the world of Financial Crime: the use of AI could help to reduce operational costs whilst also increasing the accurate detection of fraud. However, as an emerging, largely unproven technology (at scale in banking), the effective and widespread use of AI is heavily contingent on regulators accepting the banks’ approaches, and even more so where financial crime is involved.
Transaction Monitoring costs are dramatically increasing as global non-cash transaction volumes grow year on year, which has led to a large increase in the number of suspicious transactions which need to be investigated by compliance officers. Coupled with this, Transaction Monitoring has always been an area of great scrutiny for regulators as they require clear explanations for any decisions taken. Due to the low tolerance for mistakes, financial institutions have leant towards a rule-based approach which allows them to explain their decision making processes.
AI provides an opportunity to reduce the operational costs associated with Transaction Monitoring by re-designing solutions to switch from a rule-based approach to a risk-based approach. However, this will inevitably make decisions more difficult to explain – not only will the system begin to make decisions that deviate from hard coded, black and white rules, but results may not be reproduced even with the same input data. The black box effect of artificial intelligence is hotly debated in the industry, and poses something of a challenge for regulators that want to maintain control and oversight of key banking processes, but also want banks to embrace new technologies that could make the market safer for customers.
That said, it seems that regulators are actively encouraging financial institutions to utilise innovative approaches to combat money laundering. In December 2018 the US regulators released a joint statement suggesting that banks and credit unions take “innovative approaches to combating money laundering, terrorist financing and other illicit financial threats.” It detailed that pilot programmes using AI “will not necessarily result in supervisory action with respect to that program.” This means that if banks were to implement an AI solution that identified suspicious activity which wouldn’t otherwise have been detected, then the regulators won’t automatically assume that the banks existing processes are deficient. It seems that the regulators realise the need for better transaction monitoring to reduce financial crime on a global scale. They believe that innovative solutions such as AI could provide the answer and are encouraging banks to investigate these possibilities.
If the desire from both sides is there, then surely there will be swift adoption? One would expect this to be the case, however, replacing transaction monitoring systems is costly and the knowledge and skillset to implement AI projects in banks is still growing. The key to progress is collaboration between the regulator and the industry. Collaborative efforts can ensure wider adoption and help set the standards for model management of AI based transaction monitoring solutions. Increased guidance and support from the regulators will provide financial institutions with the safety net to invest in innovative technology. This in turn will help them reduce operational costs and increase compliance, killing two birds with one stone.