November 12, 2019 | Written by: Ciaran Doyle
Categorized: AI | Banking | FinTech | IBM RegTech Innovations
Share this post:
About a month ago, I attended the IBM RegTech Summit in London, which brought together a mix of financial services professionals, regulatory experts and technologists. But the terminology was markedly different than most financial crime and compliance events I’ve attended. With terms like “AI,” “machine learning,” “cloud” and “innovation,” you could make a successful run at buzzword bingo. But there are new terms that typically haven’t been heard amongst former regulators and current risk and compliance professionals. “Transparency?” “Perception?” “Openness?” Who are these people?
Openness is the only path to sustainable innovation
While the openness mantra seems suited to a Silicon Valley start-up, it’s equally fitting for the current environment of regulatory examiners and the financial institutions they oversee. The days of “black box” innovation, inexplicable neural networks and decade-old models that “just seem to work” are over. They are coming to a swift end with the increasing adoption of AI.
In the same way, the classic rift between internally-built detection models and vendor-driven analytics are starting to blur, as open source and interoperability become less fringe and more fundamental. Part of this is driven by the growing number of available tools and skill sets. Technologies are becoming more intuitive and accessible, while analytics and data science expertise – though still highly sought – is more readily available to the financial industry.
But transparency and explainability are more than product benefits of this new, more open product design mentality. As regulators encourage innovation in the area of financial crime, the demands on financial institutions to be able to easily explain and justify decisions made by their AI technologies will become essential.
Buying (and building) for the future
Part of the hesitation among banks to adopt AI in the financial crime space – more so in AML compliance, as fraud investigators have been using various AI techniques like machine learning for years – is the regulatory perception and comfort of understanding why the machine made a certain decision. Unlike traditional, more static models, AI is continuously learning and changing. Exposing both the rationale as well as the data that led to those decisions is important to show the model is acting as originally intended and there hasn’t been any bias or undue influence to skew results. Obviously, the transparency and explainability will be absolute requirements for that.
Another roadblock in the adoption of AI will be what banks have dealt with over the last two decades, which are so-called point solutions that aren’t well differentiated and only focus on single aspects of the problem. These systems are hard to adapt to changing market conditions and advances in technological development – which explains why interest in AI has received so much attention in financial crime prevention as a whole, not just fraud detection.
But what’s to stop the currently innovative AI fintechs from growing into those legacy vendors in the next ten years? The answer again is openness.
Flexibility and integration into the institution’s larger ecosystem are table stakes, but the ability to readily shift to cloud should also be a key consideration as organizations plan out their next decade of technological investments. All these different factors are converging, and making it difficult to assess whether today’s emerging AI startup or existing market leader will be able to better evolve in an uncertain future.
Bringing AI, cloud and adaptability together
The world of financial crime prevention is about managing constantly changing threats and trying to stay a step ahead of the latest fraud, money laundering or criminal scheme. Up until now, that adaptation was slow, and often required the help of outside vendors to tune, update or create new detection models. But to meet this quickening pace of change, we’ve created an approach that can get ahead of emerging threats in financial crime.
IBM Financial Crimes Insight is a suite of AI-driven solutions that solve some of the most pressing challenges in financial crime. Intelligent prioritization of AML and sanctions alerts. Real-time payment fraud prevention. Uncovering fraudulent insurance claims. Spotting potential employee conduct risk across communications channels. These capabilities help financial institutions address their current needs and the areas where efficiency gains and more effective detection are most sorely needed.
Of course, as AI skills within an institution grows, the organization will want their own data science teams working to actively thwart financial criminals. That’s why again, openness is key. The IBM commitment to openness gives financial institutions the ability to manage the embedded AI and machine learning capabilities that we provide. In addition, we enable organizations to quickly and easily incorporate their internally-developed models. This combination of ready-made solutions and the ability to continuously adapt helps banks, insurers and financial services organizations address their main financial crime challenges today as well as those that don’t yet exist.
Want to learn more about fighting financial crime with AI?
You can learn more about how we’re approaching this exciting time in our recent report “Fighting financial crime with AI” or explore some of the ways we’re addressing the wider needs of risk and compliance through our website below.