"confident but wrong" is a very reviewer 2 vibe, so we should not be surprised by the rise of llm use in reviewing paper submissions
i guess people are already doing this but i suppose best practice is to submit your own papers to the AI gods and ask what is wrong with them, because that is what reviewer 2 will do anyway; "missing the point as a service"
If you have a fediverse account, you can quote this note from your own instance. Search https://mastodon.social/users/wingo/statuses/114693038723611610 on your instance and quote it. (Note that quoting is not supported in Mastodon.)