A recent World Economic Forum survey found that 85% of financial institutions have implemented AI, but some still have concerns about its use. To help financial institutions apply AI responsibly, Oracle has added new AI governance capabilities to its Oracle Financial Services Compliance Studio application. With a new visualization canvas that fosters more inclusive machine learning model creation, alerts for potentially sensitive issues, and new in-memory sandboxes to more easily create and refine models, banks can help weed out bias and uncertainty in applying the power of AI to fight money laundering and other financial crimes. These new model management and governance capabilities will also soon be implemented with PwC’s Model Edge solution.
Oracle Financial Services Compliance Studio is an advanced analytics application for financial institutions that supercharges anti-money laundering and anti-financial crime programs for more effective, efficient customer due diligence, transaction monitoring, and investigations. As part of the broader Oracle Financial Crime and Compliance Management suite, it offers built in AI, machine learning, graph analytics, and data management capabilities.
“There is clear consensus across financial institutions that AI and machine learning have tremendous potential to increase the effectiveness of anti-money laundering and other financial crime detection programs, and deliver higher efficiencies in the investigative process,” said John Edison, group vice president, software development, Oracle Financial Services. “However, there are hurdles to ensuring that AI is used responsibly and ethically in a way that is fair, transparent, and easily understood. The new capabilities in Oracle Financial Services Compliance Studio provide the right governance and controls required to meet these objectives.”
Current compliance processes for adopting AI within financial crime are often disconnected, error prone due to manual steps, and resource-intensive. With the new updates to Compliance Studio, banks can eliminate these obstacles to better address their governance and regulatory obligations. New features include:
A visual canvas for building machine learning models: A visual, drag and drop interface provides a task library so data scientists can quickly and easily build machine learning models that reflect company values and help address regulatory guidelines. The pre-trained models and data sets also allow non-data scientists to more easily analyze and consume the machine learning insights. The visual representations of models are easy to interpret and explain to regulators.
- Interactive model testing: The visual canvas of Oracle Financial Services Compliance Studio allows users to interactively test and edit models, making it easier to pinpoint and help address issues of bias or fairness. The visual canvas enhances decision-making by providing context in an easy-to-understand format.
- Model monitoring and alerts: Active model monitoring and drift reporting provide immediate alerts on areas of concern so corrective action can be taken.
- Easy to create sandboxes for business users and citizen modelers: Business users and administrators can create and use multiple in-memory sandboxes to bring data from multiple sources and test models without disturbing the production environment.
- Fast point and click model deployment: New and updated models can be deployed straight from the application in minutes without taking systems offline.
“While many financial institutions are already using artificial intelligence to help detect money laundering and fraud, the management and governance of AI models tends to be manual and cumbersome, which leads to errors and inefficiency,” said Vikas Agarwal, Financial Crime Unit and Risk Products Leader at PwC. “Financial institutions need technology that helps make this ongoing process seamless and that provides a holistic, quantifiable picture of model risk through various dimensions. We look forward to working with Oracle and combining their model management and governance capabilities with our PwC product, Model Edge. Bringing this technology combination to the market will help ensure AI is used responsibly, documented well per model governance guidelines, and that any potential biases in models are identified and addressed.”