I think this was brought up somewhere in a sibling thread. It would be great to have a more granular content restrictions framework that allowed for exactly this. But how to do this in a reliable, verifiable way sounds like a real challenge.