How is Generative AI impacting the KYC process?
Ever since its launch in November 2022, Generative AI has been the topic of discussion on everyone’s lips when it comes to discussing transformational technology. In the area of KYC, could we expect Generative AI to have significant long term impacts?
In the view of Ted Datta, senior director, head of financial crime compliance practice, Europe, Africa & Americas at Moody’s, Gen AI can make a significant impact on KYC processes, but only when it is trained on trusted and valuable data.
He added, “One area where this applies aptly is in the creation of riskographies i.e. risk-relevant databases of financial crime information. Using Generative AI, Moody’s recently created ten times the number of new riskographies for a client in only 38 days, allowing the client to review each risk alert in significantly less time.”
While GenAI led to the quick development of these riskographies, Datta exclaims, the increased efficency is counterproductive if the information it presents doesn’t paint an accurate picture.
He continued, “Every model and prompt integrated into compliance products must have a complete governance framework, not only to assess accuracy but to ensure that industry-standard observability and explainability of the models are maintained.
“The prime activity for many compliance professionals is to understand who their organisation is doing business with, so any enhancements to KYC processes from generative AI need to have the highest accuracy, something that’s only possible with Generative AI relying on curated data sets.”
GenAI benefits
Among some of the most prevalent benefits of GenAI are its ability to detect, mitigate and prevent Financial Crime throughout the KYC journey.
“It is capable of generating insights and summaries about customers, thereby reducing the manual workload for FinCrime Investigators, ensuring consistency and allowing them to dig deeper and spend more time on complex cases,” explained Laura Hadfield, anti-money laundering SME at Qkvin.
She notes, however, as it provides FinCrime professionals new approaches to mitigate these risks, so to can malicious actors leverage these new approaches to stay ahead of detection.
“GenAI can be used to produce audio, images, videos, and more, all of which can be used in KYC as part of the identification process. The use of GenAI for deepfakes is particularly concerning where it could evolve to bypass liveness detection.
“As GenAI becomes more sophisticated, such as incorporating environmental factors and realistic lighting, traditional fraud detection methods will need to rapidly enhance to distinguish between GenAI and genuine media, as part of the KYC process. Equally AI can also be used against AI to detect such deepfakes and fraudulent assets.”
Hadfield also highlighted presciently that industry will be forced to continue to review and update risk management frameworks, systems and procedures to resist novel methods of bad actors utilising GenAI to protect the integrity of KYC processes.
