Potential alternative to this: limit interactions based on label i.e, you want to have a public discussion about AI, but you don't want people popping in to indiscriminately hate on it, so you limit replies based on the "AI Hater" label.
Community-spaces on Bluesky will solve this. AI isn’t a subject matter fit for the town square. It can be discussed publicly, but the audience has to be a self-selected, softly bounded public of technologists with shared context.