Doesn't having @cheetah post a comment about (potential) plagiarism actually help human readers and commenters know when they are interacting with a potentially bogus post?

Maybe if there was a way to have comments that are purely to add metadata about posts like additional community added labels, or quality metrics, then bots could be restricted to adding only those.

Another thing that can be done is to actively encourage downvoting of posts purely for low quality and issues like that - instead of reserving it for abuse and spam. Some people call that "abuse of the rewards pool" anyway. I'm actually more inclined to have Steemit just rename "flag" back to "downvote" again and then lobby for metadata labels that can be voted on and be used to appropriately filter what Steemit.com and other Steem UIs show.