Reframes AI data-leak exposure as a board-level metric, using 2025 breach data to argue for a control layer at the AI traffic boundary.
