AI’s role in finance increases—and so do concerns

The financial services industry is trying to balance its enthusiasm for AI with concerns about security.

While the technology can streamline core processes for the industry, it also needs parameters, either at the regulatory level or with safeguards in the tech stack.

Talk to me. One of the ways companies are deploying AI to their operations is to add in ways for apps powered by the technology to communicate important data to stakeholders, said Danny Kang, CEO of Ducenta Squared Asset Management. His firm uses the technology to translate information about the markets to customers more quickly.

Ducenta, founded in 2020, has $4 billion in assets under management, meaning that the company is deploying this AI-based communication to optimize its portfolio. The AI is effective and secure for customers, Kang told IT Brew, because there’s a clear and workable regulatory and internal framework managing how it’s used.

“Data integrity is number one, the biggest issue,” Kang said. “Number two…is where that data is being housed, and who’s able to touch that data in relation to our obligation for regulatory adherence.”

Lock down. The tension between speedy integration and effective security is seen across the fintech space. In addition to large companies like AWS’s technology, there are smaller firms like Hapax, an AI platform that’s offering credit unions and banks the ability to deploy agents throughout their systems with varying degrees of human interaction.

Because it works with 20 financial institutions managing between $250 million to $90 billion in assets, according to company spokesperson Kevin Green, Hapax’s platform needs to be secure. While it’s critical for financial service companies to control how they apply AI, the technology has its share of vulnerabilities.

Hapax co-founder and CSO Connor Huddleston acknowledged there’s potential for the product’s permissions to open the door to an expansion of the threat surface, but trusts his clients’ IT teams will follow best practices to secure their environment, and control access as a default. There are protections built in, he added, that flag any irregular behavior from the agents.

“We don’t straight-up prevent it, unless that company has set some sort of human-in-the-loop approval required for it, or a guardrail around it,” Huddleston said. “But out of the box, it’s not going to go do that unless you just remove all restrictions.”

Pros and cons. To Kang, AI integration into finance is dependent on its usefulness. If it’s not convenient or easy to use, it becomes less necessary, and IT teams will feel less urgency to deploy the technology.

“If it assists in terms of communicating or delivering the primary goal of the company and to its stakeholders—whether it’s clients, consultants, developing corporate behavior where the goal is simplified—then I welcome it,” Kang said. “But if the AI ends up complicating things, I shy away from it because…that means that there’s a staff person that’s missing in the room that can help facilitate simplifying certain things.”