I’d actually like it if their content moderation AI was trained on users content. Specifically content reported and actioned. So that the tooling becomes more intelligent and less reliant on content being reported in the first place.