There’s Room for Generative AI in Financial Services

Cohen argues that, from a broad perspective, the general principles of current codes of ethics and other policies already apply to the use of AI.

“You cannot share material non-public information with others,” says Cohen. “That would include putting it into an AI or similar system that's publicly available or available to the provider.”

While there’s some existing policies that can be applied, there’s still a need for more specificities to move forward with AI. While prohibition was the prevailing policy, there was also a rollout process that educated employees on what fell inside and outside the scope of that prohibition rule — which is how many people were able to initially implement AI use.

Now, people have a much better handle of the implications of AI and how their firm needs to use it. For any firm that has been on the prohibition side of AI, the first step is to lean into more flexibility around AI use. Cohen advises firms to gain an understanding of the different groups and stakeholders, and how they would like to use AI.

At the end of the day, if you’re going to move forward with AI, your people need to gain internal expertise on AI capabilities and implications, even if you’re relying on external assistance for oversight and supervision.

This all comes back to off-channel communications.

Many firms have already established and expanded processes to deal with off-channel communications. If a firm has Copilot and Teams but doesn't have a policy defining how they can be used, that’s a potential off-channel exposure to address. The work that's been done over the past two years to get those processes in place now can be extended and leveraged for AI — and how it’s going to be evaluated.