text only - images, videos, and other binary content must be stored elsewhere (could still have links though)
data is stored encrypted (user option) - wouldn’t show up in scans
clear separation between “stuff I’m interested in” and “random stuff I’m hosting for the network” - plausible deniability
perhaps an option to opt out of random data storage - you’d still store stuff you’re interested in
But yeah, that’s a huge part of why I’m unwilling to share the code until I’m comfortable with the moderation engine. I think there’s an opportunity to use that moderation engine to reduce storage of CSAM (i.e. don’t store data from blocked users).