I hardly need to point out the ways in which Artificial Intelligence (AI) has exploded into our lives – both personal and professional – over the past few years. I recently read an EY report that said that 90% of European financial services firms have integrated AI to some extent and 69% expect Generative AI (GenAI) to significantly impact productivity. To take an example from our own industry, GenAI is often put forward as a way to streamline the more labor-intensive and time-consuming tasks in KYC operations.

The way I see things, there’s no denying that AI holds gamechanging potential. However, I think it would be foolhardy for executives not to acknowledge its risks, many of which are very complex. One of the reasons for this is that modern AI and machine learning models tend to work within a ‘black box’, with data points working together in ways that are impossible for a human to understand. So executive teams are often left scratching their heads, feeling sure that AI could offer value in their KYC processes but equally not wanting to expose their organisation to unknown risks. 

‍

To shed some light on the matter, I’ve come up with five key points for leaders considering AI in KYC to discuss with their teams :

1. How can AI enhance our KYC processes? 

Rather than getting sidetracked by shiny new technologies, we advise our customers to think about the problem they’re trying to solve, and how (or if) AI can help.

There are a myriad of ways in which AI can be effectively deployed to boost efficiencies within the KYC space. For example, AI can extract tasks and related due dates from an incoming email, generating and assigning sub-tasks for individuals to act upon. It can prompt the collection of data from counterparties – or even source the correct information from public sources. At Avallone, our mantra is that AI should act as a user’s “wingperson”. It should augment, rather than replace, the work of the human. By taking each KYC task case by case, we work with our customers to determine where it makes most sense for AI to streamline their processes. 

‍

2. Where can we find experts in AI for KYC? 

A significant portion of firms say that they have limited GenAI expertise within their workforce, and we know that Machine Learning and AI are typically areas with the widest talent gaps. While it’s important for leadership to take a long-term approach to building key skills within their teams, executives often view this issue the wrong way around. In fact, you don’t need AI experts to determine how best to improve your KYC processes. AI is a tool like any other, and what’s most important is to place that tool in the hands of someone who fully understands the problem you’re trying to solve. You need KYC experts who understand all the complexities within this space, and who also have an overview on where automation could be implemented for the greatest gain and least risk. 

‍

3. What are the potential pitfalls in using AI for KYC? 

Treasury and compliance leaders are right to proceed with caution when using AI. As AI becomes more commonplace within the industry, it’s naturally falling under increased scrutiny by regulators. For example, the European Commission’s recent AI Act promotes principles of safety, ethics, transparency and accountability - requiring transparent model decisioning, explainability and tracking of data privacy. Regulators are penalizing organizations who fail to adopt the correct approach – as they should. Unfortunately, an increasing number of organizations are falling foul to these regulatory requirements. Recently, a Danish bank used AI to close thousands of alerts, however they were unable to explain the AI-driven decision-making process and were forced to re-process them manually. In Germany, a bank was fined €300,000 for not being able to justify why AI rejected a customer’s credit card application. Failure to ensure transparent decision-making can cost an organization in time, money and reputation – often negating the original benefits of AI.

Given the significant risks that AI introduces, senior leaders must be sure that there are adequate controls in place to protect both their business and customers - often this involves balancing automation with human oversight. As well as ensuring explainability of all decisions, organizations must also stay updated with the ever-evolving technological and regulatory developments. Executives need to  ensure that they have adequate resources to manage this extra workload either within their own teams, or through collaborating with a trusted partner. 

‍

4. How - and why - should we balance automation with human oversight? 

It’s widely acknowledged that AI can perform many tasks faster, more efficiently and more cost-effectively than humans. However, particularly in KYC, human verification is a crucial step. Let’s take a practical example: collecting key financial information from investors to send to your bank. Automating parts of this process can save teams significant time and resources - AI can scan a questionnaire, and source and add missing data points. However, automatically sending this data onto your bank is a step too far. With no human oversight, your organization is left open to data breaches, errors, fines, and untold reputational damage. Not to mention the issues surrounding accountability – who is actually responsible if AI makes your investors’ sensitive information public? None of these consequences are worth the risk. Simply by building in a layer of human approval before sending the package to your bank, you mitigate against these harmful scenarios. 

‍

5. How can we integrate AI with our existing and future workflows? 

At Avallone, we focus on collaborating with our users to understand where it makes most sense to automate. We recommend starting with the obvious, repeatable tasks and building from there. By automating in small and digestible ways, Treasury teams are able to retain full control and accountability while also enjoying AI’s full benefits. And while each AI use case may seem minor, the ultimate impact on the organization’s productivity is impressive. 

I can see a future in which organizations manage their KYC work seamlessly via a web of workflows with AI powering certain tasks, and human verification used when needed. This will allow their ‘human’ teams more time for work that brings greater value to the business. And here’s the important part - the workflows must be flexible, enabling the business to select the right software partner for each of its needs. That way, the organization can scale while also always ensuring maximum efficiency.