Okay, let me create an account on mastodon.social and use it to scrape content from every other instance.
Better yet, let me create an account on “i-want-privacy-in-a-public-internet.example.com” and access the federated timeline directly, then I can go and push the content from everyone into this discovery service.
What are they going to do? Unless they go to the point of asking for physical evidence behind the person asking for accounts and/or only give invitations to people they already know, and *completely shut down their own servers to the outside world, they will never be able to avoid data leakage.
And if they do get to do any of this, then what is the point of using anything based on ActivityPub? They will be better off by just using any of the existing group chat servers like Discord (or Matrix/XMPP if they still care about FOSS.)