So far I have managed to get it down to one bot scanning for multiple
hallucinations once per day, instead of several scanning several thousand
times per day.

On my server deb.devuan.org/robots.txt returns -

User-agent: *
Disallow: /

Which seemed to stop most things. Claude and semrush I stop at the
firewall with the IP blocks I have discovered so far.

That robots.txt also stops legit search engines. Though I think the most
popular search engines are also indiscriminately using AI scanning bots
anyway.

Or maybe I've just got lucky the last two days.