Moderation on the Feviderse is different that commercial platforms because it’s context-dependent instead of rules-dependent. That means that a user accout (bot or otherwise) that does not contribute to the spirit of a community will not be welcomed.

There is largely no incentive to run an LLM that is a constructive member of a community, bots are built to push an agenda, product or exhibit generally disruptive behavior. Those things are unwelcome in spaces built for discussion. So mods/admins don’t need to know “how to identify a bot”, they need to know "how to identify unwanted behavior".