Artificial intelligence use in compliance and operations at industry firms remains a “mile wide and an inch deep,” according to representatives from the compliance consulting firm ACA Group.
While about 84% of respondents to a recently released ACA Group survey reported using AI, they also reported that “active AI use” occurred, on average, in only two of 20 compliance and operations sub-functions. Only 18% of firms that responded said they used AI in compliance tasks, compared with 5% in operations.
According to the ACA Group, the report cataloged responses from 201 investment management firms, including asset and wealth managers, hedge funds, private markets firms and broker/dealers.
In a discussion about the report last week, Joseph Kochansky, ACA’s head of product and engineering, used “mile wide, inch deep” to describe AI adoption among firms, saying it was only being done “in a very thin way” across organizations.
The most common use of AI, at 35%, was in the so-called “compliance program administration” sub-function, which includes day-to-day tasks such as summarizing reports, preparing first drafts of communications and reviewing policies and disclosures.
However, ACA Global Regulatory Compliance Practice Co-Head Josh Broaded noted this sub-function captured users working with desktop AI tools, including Claude, Microsoft Copilot and ChatGPT. During last week’s discussion, Kochansky acknowledged that while desktop AI was easy for users, it can be more cumbersome for compliance management.
“It’s not integrated into your workflow in a way that you can get the full benefit,” he said. “So while compliance program management was the biggest use case, the reality is … that desktop use case was really the dominant use case, and it’s driving a lot of that usage.”
After compliance program administration, ACA Group found electronic communication surveillance to be the next most common AI use case at 30%, followed by marketing material reviews (28%), compliance testing (22%) and employee compliance monitoring (18%).
Compliance testing and monitoring also topped respondents’ “AI wishlist,” with 52% planning to incorporate AI there next (or more than they currently have). While 21% and 20% of respondents respectively marked e-comms surveillance and marketing/advertising reviews as their next targeted use case, 17% wanted AI incorporated to automate repetitive tasks and for data analytics/reporting.
According to the survey, operations remains the “untapped frontier” for AI adoption, with minimal average usage across the category’s sub-functions. At the top of those sub-functions is market data quality control, but even it had only 9% of respondents currently using AI. Cash and position reconciliation followed at 6%, with trade confirmation at 5%.
Among the speakers discussing the report last week, the “age of agents” or agentic AI, will usher in both new opportunities and complications in reference to adoption in the financial services industry and within compliance/operations divisions (unlike generative AI tools like ChatGPT, agentic AI tools can autonomously perform functions in pursuit of a goal).
“In many ways, you’re embedding your application into AI rather than embedding AI into your application,” Kochansky said. “And that creates a lot of the power, because AI itself can now iterate and try different approaches to solve the problem that is at hand.”
