Exactly

I don't understand it too, e2e encryption but the AI has to process the data somehow :L ???

EDIT: If they refer to encryption as just encryption on message transport that's just HTTPs, to verify that they are not saving your data, the backend source code needs to be available to analyze it.

If your encryption is not a layer on top of a messaging service, you have to trust that the service you're using is actually end-to-end encrypted. I point this out because it means that encryption is not a protection against he service not doing what it says it does, but rather it is a protection against other things: passing data to governments, having a hacker break in and leak it, that kind of thing.

By storing stuff securely, it mitigates that problem, I guess. A government would have to have a "live tap" to know what you write to the LLM, rather than being able to slurp out all your historical conversations.

I didn't see mention of it, but maybe they could use homomorphic encryption?

in systems i've seen with severe encryption, the data is only unencrypted by the actual processing engine. otherwise, its encrypted at the field level, during transit and at rest. yes, you have to trust the processing engine isnt doing anything nefarious, but at some level you have to trust something. proton has a solid history of trust despite the one idiot that works there.

youre not smelting your own processors for example. if youre not forging your own processors then youre just not secure!

I came so close to switching over to them at the start of the year. I was in the middle of planning that change, then their founder or CEO or whoever he is started going political. So I held off. And now just like everyone else they've jumped on the AI bandwagon.

Because people are mostly incapable of using the button as anything other than "I like this" or "I don't like this".

there are a metric tonne of ai haters. it doesnt matter what protection are put in place; their butt huuuuurts.