AI is helping banks supercharge their ability to detect financial crime, adding context to sanctions lists, cutting down on false flags, and speeding up investigations.

Artificial intelligence is bringing more context to compliance, according to Ryan Schmiedl, chief data officer and head of trust and safety for JPMorgan’s payments division.  

“With the innovation going on with large language models, there’s a lot of opportunity in the financial crime space,” he said on stage during the Finovate conference in San Francisco. He gave an example of how LLMs could help algorithms better understand the context of a sanctions list:  

“With sanctions, most of the time you get a list with very little context to what you’re looking for other than a name, or vessel number, or particular company. And so you’re searching through a bunch of unstructured text,” he said. “In the old systems, if vessel number 138 was flagged, guess what? Every time you had ‘138 Main Street’ in the data, it’s a hit, right?” The latest algorithms can use context to parse out what’s actually meaningful and cut away “a lot of unnecessary friction” and false flags, Schmiedl said.  

Ultimately, the goal is to use AI to catch more fraud, faster: “There’s value in offloading very manual, repetitive tasks that don’t require a lot of human IQ, that have a basically consistent pattern,” he said. “We’re laser-focused on capture rate and precision: We want to capture a larger percentage of these malicious events, with better accuracy and fewer false positives.” 

Insights Distilled has previously covered the various ways that FinServs are using artificial intelligence to fight fraud, including through better know-your-customer checks and more comprehensive investigation reporting.