Mike Hsu

On March 20, 2025, at the invitation of Claudia Buch, Chair of the ECB Supervisory Board, I made a presentation on “Cybernetic Supervision” to the full Supervisory Board in Frankfurt. I framed supervision as a feedback mechanism (see graphs below), where the challenges of minimizing both “type 1 and type 2 errors” are heightened, especially given the complexity and growth of the banking system and faster pace of information and money flows. I proposed a solution: cybernetic supervision — where supervisors are much more agile, empowered, and aligned –assisted by AI and changes in processes, priorities, and approaches.

The graphics below are excerpted from the presentation. They show supervision as a feedback mechanism, identifying and informing banks of unsafe and unsound practices.

The failure of supervisors to identify or address unsafe and unsound practices or conditions, as they did in the lead up to the 2008 Global Financial Crisis, constitutes a “type 2 error” (false negative).

Conversely, the incorrect identification of certain practices or conditions as unsafe and unsound, when in fact they are not, constitutes a “type 1 error” (false positive).

Type 2 errors can lead to disorderly bank failures eroding trust in banking and regulators, while type 1 errors erode the credibility of regulators and misallocate valuable resources. As the banking system expands in size and complexity and the speed of innovation, information, and money increases, the risks of both type 1 and 2 errors grow. Cybernetic supervision can help supervisors meet this challenge.