Boards can't own AI risk they can't see
The Financial Stability Board's final guidance puts boards at the center of AI risk in finance, and that's right. But a board can only own what it can see. On the AI approval committee I served on, the hard part was never the policy. It was keeping an honest inventory of where AI was actually running and rating each use case's materiality, then revisiting that rating after launch. The FSB asks for assessment at the outset and on a continuing basis. Most programs I've seen do the first and quietly skip the second. If your board got a list of every AI use case tomorrow, would anyone trust it?
Source: FSB puts boards on the hook for AI risk in finance - FinTech Global (FinTech Global)