AI can’t on its own prove that all the appropriate safeguards are in place or that all CUI is in the right place.