I have some aggressive filters in Caddy that block the worst offenders by CIDR range, and also filter by user agent to remove any honest facebook and amazon bots. Otherwise, maybe strong rate limits by IP?

Edit:

Longer term, the approach might be - provide a separate RSS feed with full content but gated by a query parameter, then only give that URL to known-good consumers via email verification or patreon subscription, etc.

It would suck that people would have to pay more to consume content in their preferred way, but depending on your needs it might be a reasonable compromise.