With interaction I suspect of being inauthentic (follow, clap, meaningless comment, highlight), I now take a quick look at their profile, and if I suspect them of using AI for their articles (titles, frequency, cover images and other evidence), click into one, make a quick decision as to whether it's AI in <30 seconds, and then report for AI use and block writer.

I'm running out of patience doing Medium's work for them, and have been blocking and reporting 15-20/day for the last couple of weeks.

They must have enough data now to automate the process, if that's feasible. If not, the site is in a bad way.