Battling with bots: Fighting financial crime in the AI era

Money launderers are an innovative lot. So it is not surprising that the perpetrators of financial crime have been quick to adopt AI. They are using it to write better emails, to develop better code and to fake access codes. They are applying generative tools and using unsupervised learning models to exploit security gaps. They are using prompt engineering to circumvent controls.

The good news is that financial institutions are also rather innovative. And we are seeing significant uptake of new AI tools and technologies across the financial crime detection, prevention and management world. Banks are using AI to improve detection and reduce alert volumes. They are using it as a first line of defence for data security. They are using it to scan millions of transactions every day to spot anomalies. Algo a algo; it’s exciting stuff.

Yet the real financial crime-fighting AI heroes in today’s leading banks and financial institutions aren’t the front-line bots. It’s the boring back-office ones. The reality is that KYC and AML record-keeping and data management soaks up the vast majority of a bank’s total financial crime resources. Reducing this burden, therefore, would release capacity to focus on more value-added activities like investigations, proactive strategy development and controls improvement.

Recent reports suggest a number of very big global firms are already seeing massive benefits from applying AI to their KYC and AML processes. One major firm claims that their AI models have made their KYC processes 90 percent faster, allowing them to process the same number of files with 20 percent less people. Another major global bank claims to have reduced their number of false positives by 60 percent.

I recently participated in a panel discussion hosted by Risk.net’s Philip Harding alongside Guy Mettrick, Industry vice-president, financial services at Appian (a leading process automation solution) and Jay Krish, Head of data governance for financial crimes compliance at State Street. You can watch the full discussion here.

What came out of that discussion was an interesting list of topics that financial crime leaders should be considering as they look to apply AI to their operating models. Here are the top three:

1. Connecting your data: Clients often complain that their data quality is poor. But, more often than not, the real problem is that they are asking too much from a single source of data. What they should be doing is focusing on using technologies to connect different data sources together in ways that unlock insights and identify risks. This isn’t about having a perfect pool of data, but rather how you combine your data to get reliable insights.
2. Improving your processes: We see organisations spending a lot of money on new technologies only so they can do the wrong things more efficiently. As with any occasion when a new technology enters the workflow, it is worth re-examining the processes you are enabling to ensure they are effective, compliant and robust. AI isn’t just an opportunity to turbo-charge your processes and increase productivity – it is also an opportunity to fundamentally reimagine your processes for the future. 
3. Monitoring the risks: The pace of technological change makes it very difficult for financial institutions to stay ahead of the more innovative criminals. The reality is that most of the big financial crime events have happened in those ‘grey’ areas where emerging technologies create unknown vulnerabilities. The challenge, therefore, is to stay vigilant and to collaborate with advisors and solution providers who can help you and your organisation stay ahead of emerging financial crime risks.

In the webinar, we also spend time sharing some of the best practices we’ve seen for leveraging technology to enhance entity resolution, screening and payment monitoring while aligning with regulatory requirements and optimising operational efficiency. And we go into more detail on how advanced AI can help streamline AML, KYC and fraud detection.

If you are interested in financial crime management, I would highly recommend watching the replay. Or contact KPMG to learn more.