Seems like just another reason why defederation should be completely removed from the protocol. It’s way too easy to abuse and force centralisation.
There are other far less destructive and abusable ways of dealing with spam and content moderation.
I maintain that it’s better to give the users the control, and allow them to decide which instances, communities, and users they want to be exposed to. Bottom up moderation, instead of top down.
For example, instances can provide suggested ‘block’ lists (much like how an ad blocker works) and users can decide whether or not to apply those lists at their own discretion.
By forcing federation, the network stays decentralized. Maintaining community blacklists that can be turned on or off by the individual user protects against heavy handed moderation and censorship, whilst also protecting users from being exposed to undesirable content.
Honestly, training their proprietary AI on GPL code (with no ability to opt out) was enough for me.