If someone is selling drugs, distributing copyrighted material, publishing defamatory things, or participating in any other crime on your platform and you know about it, you lose 230 immunity for all content. Maybe this would encourage E2E encryption, but... If you read further, every single provider that allows user content also has to have a tipline for criminal activity and keep records on it.
Having a tipline isn't so bad on its own, but then the problem becomes: User leaves unverifiable tip, what do you do with it? It's not like you can break the encryption...
> Sounds like we are turning our entire website infrastructure into a 1984 monitoring system
It already is one for all intents purposes, just not one that is used ubiquitously (except in the sense of getting us to buy stuff). All that is changing is that the panopticon is becoming slightly more explicit than implicit, making it that much easier to (eventually) flip a policy switch that designates groups wholesale as "enemies of the people" requiring active scrutiny and interference.
Ingsoc didn't monitor proles, and didn't always monitor inner party members, so I'd claim its monitoring reach was only somewhere between half and a quarter of our entire website infrastructure.
If we’re talking about content that you literally don’t have access to and can’t see because it’s encrypted, like iMessage or Signal then you don’t need Section 230 protection because you aren’t moderating content.
A statement from the user who submitted it, including their unverifiable copy of the offensive message. The law is not computer code. They don't need to break ciphertext to get a warrant.
You can do this thing called "message franking" to make it more verifiable, if you want.
The downside is, context matters. If you report messages out-of-order or with important context deleted, you can trick investigators into thinking something was said or implied that actually wasn't. It needs to be carefully designed.
(I blog about cryptography, but you should ask a cryptographer if you want to design something like this.)
I don't design message apps. I just know that law enforcement did just fine before the internet when most communication (face to face, over the phone) wasn't recorded or verifiable.
