Active moderation isn’t spying but using an AI is? The only reason those self-hosted community servers didn’t have problems was because they (usually) had active admins to see bad behavior and take action. This is merely automating that so a real human being doesn’t have to be there watching.