Spent a bit of time during lunch outlining a cryptographically secure (???maybe???) sensitive content tagging system that is both plausibly deniable, and balances protecting consumers + creatives: https://codeberg.org/conjured/cauldron/issues/2
You can't search FOR something, but stuff can get filtered OUT (quietly)
https://www.schneier.com/wp-content/uploads/2016/02/paper-clueless-agents.pdf
One imagined use case for that is for a "blind search" of a patent database, which is where you want to know if a patent exists but don't want to reveal your idea for a similar patent.
I'm not exactly sure what you're trying to achieve, tbh, but it seems that there might be some overlap there.
It's important to help folks avoid works that will genuinely leave them sobbing in a corner all weekend with flashbacks. That's legitimate!
But in this day and age, having obvious, visible content warnings means that malicious actors can just search for them and get works pulled down.
Agency! yay!
