All the impersonating responses I've received (not just to responses I've made to your articles, but quite a few others) were a response to my response - so two levels away from the article.

Medium doesn't need to employ AI to deal with this - it has *readers*. We just need an additional box to check when reporting a post.

As it is now, you tap the three dots, then report, then you get a choice of reasons. I've been picking "Spam", but that's not exactly right. If only there was one labeled "Author impersonator" or "Impersonation".

There'd still need to be a human in the loop before the offending account was banned, but just looking two steps up to see if the impersonating author had a name hard to distinguish from that author should make it straightforward - and hard to 'game'.