AI And Regulatory Compliance For Financial Services
Join The Discussion | Ai4 Finance 2019
Regulation and regulatory compliance is a highly-necessary component of the continued success of the financial services industry. The unfortunate reality is that the costs of maintaining a compliant business can be staggering. With legislation changing daily, the complexity of the regulatory landscape is only increasing. RegTech is the subset of FinTech that deals with technologies specific to keeping banks compliant.
There are more than 750 global regulatory bodies with 2,500 compliance rule books, producing an average of 201 daily regulatory alerts. Point being, RegTech has become a hot space, estimated to reach $6.5 billion by 2020.
Currently, it’s estimated that financial services institutions already spend ~$270 billion on obligations related to compliance. On top of that, banks have had to pay $342 billion in non-compliance fines between 2009 and 2017, with individual fines as large as $1B each.
Staying up-to-date with the changes is only one component of the problem. Data management is the core source of problems. Lapses in reporting are typically attributed to issues with data capture and storage associated with the failure to provide end-to-end data lineage.
As big data and AI technology enters the equation, there is hope in building a much more streamlined system where real time data analysis makes periodic reporting a thing of the past.
Here are some examples of how AI is being applied to regulatory compliance within financial services:
- Keeping up with regulatory changes: Banks can leverage AI to help Identify global regulatory and compliance updates, and then compare the new requirements to current bank policy in order to identify in real-time internal regulatory gaps. Such a system can go a long way in helping banks avoid non-compliance fines, while massively cutting the costs associated with having humans monitor regulatory changes.
- SEC uses AI to detect investment advisor misconduct: By applying natural language processing, the SEC is able to discover patterns in SEC filings. They are then able to use machine learning models to compare these patterns to past outcomes and determine if a filing is likely to represent misconduct. While not perfect, the SEC says that this method is 5x better than random at detecting potential misconduct.