News for hackers. Stuff that matters.
Re:Is Telegram really an encrypted messaging app? by foresto on Sunday August 25, 2024 @08:52PM (#41352929)
Metadata is indeed an open issue on Matrix. I believe addressing it is on their to-do list.

Many rooms are not encrypted because they are public rooms, where there would be no point in it. Encryption has been the default for quite a while now.

Re:Is Telegram really an encrypted messaging app? by _flux on Sunday August 25, 2024 @11:17PM (#41353706)
> I believe addressing it is on their to-do list.

I doubt that it's very high on that list, as the problem seems a very hard. Very hard as in that do we even know it's possible? "Metadata" includes a lot of stuff, but basically the originator, the destination and the timing of the messages and participants of a room are all quite difficult to hide in a federated system.

I do believe there is a plan for getting rid of the association of one user in multiple rooms, but that's but a small bit of metadata. I think it is part of the puzzle for supporting changing homeservers.

Re:Is Telegram really an encrypted messaging app? by foresto on Monday August 26, 2024 @12:02AM (#41353908)
I was referring to the metadata that are typical complaints about Matrix, like usernames and reactions.

> "Metadata" includes a lot of stuff, but basically the originator, the destination and the timing of the messages

Indeed. AFAIK, sender/recipient correlation cannot actually be protected at the software level, because packet switched networking necessarily reveals it. The common way I'm aware of to mitigate this problem is at the network level, by trying to avoid common routes that would allow monitoring many users' traffic from any one place.

Concretely, that might mean having everyone use Tor (which some folks suggest already) or going fully peer-to-peer (which some messengers do already, and Matrix has been experimenting with). Signal tries to improve the situation with Sealed Sender, but I'm pretty confident that can't protect against the Signal servers being compromised.

Related:

https://www.ndss-symposium.org/ndss-paper/improving-signals-...