FinTechTalk: How ready are financial institutions for AI-driven compliance 

Please register or log-in to watch this on-demand episode

 

Where is AI used in financial compliance? 

AI can have an impact in different areas of compliance. It can automate repetitive tasks such as sending out reminders, some routine aspects of reporting or transaction monitoring. Why regulators still ask for what they always have, their concerns about the use of AI in financial products is increasing, particularly the ones sold to retail customers. Algorithmic-based financial products have already been around for some time, and regulators always wanted to know whether who deploy AI know what it is doing and can also demonstrate their understanding. The fundamental question about regtech is if it can reduce compliance risk, cut costs or both. Another important area of focus is compliance programme operations, focussing on how all the ongoing activities defined by policies and procedures are being monitored – and here digital technology and AI play an important role too. However, AI lends itself more readily to deployment in certain areas than others. For example, it’s widely used in transaction monitoring in AML, KYC refresh, trade surveillance and adverse media screening.  

 

The major decision financial institutions must make is whether they want to procure the technology from vendors or build it in-house – in both cases robust governance in line with the jurisdiction’s AI regulations – ensuring explainability and auditability too – is key. But compliance is a special field requiring specialist technology.  

 

How sandboxes and central data platforms can enable compliance 

Traditional, rule-based ML is well suited to a slew of compliance tasks, while probabilistic AI can be leveraged to manage unstructured data. Highly regulated financial institutions must adopt a risk-based and multi-phased approach to AI deployment, documenting outcomes and decisioning. Regulators often provide sandboxes for them – an environment where they can experiment with their AI in a safe and secure environment. AI shouldn’t be an isolated piece in the governance framework but integrated into it, while compliance officers – or, ideally, even the C-suite – need to get educated about AI and its use. A full service platform like MyComplianceOffice, ensures that all data – cleaned and complete – is pulled together into a single source of truth and is ready for AI. Where AI can make a transformative change is in compliance programme management through simulating how doing business in new legislations with a different business context and risk appetite will impact a business’s policies and procedures. While humans find the complexity of this task overwhelming, AI has all the capabilities to deal with it.  

 

 

The panel’s advice 

• As a compliance officer, the first question you should ask is whether your company uses AI in the delivery of its financial products.  
• Your policies and procedures are at the heart of your compliance programme.  
• The Mills Review by the FCA aims to enhance AI adoption by integrating AI regulation into existing legislation rather than hamstringing innovation.  
• As probabilistic models, LLMs will never deliver the 100% sure outcomes that regulators expect, so it will remain crucial that humans stay in the loop – only maybe less so as AI technology improves.  
• The new AI aspect of compliance is making the role more appealing to young talent, who can get the breath and variety in this function that others can’t offer.