DMs are basically comments without a post and only one regular user can see it. No encryption, and it can be viewed by admins. (Private message reporting is a thing)
DMs on the fediverse (and Lemmy) are posts with a specific visibility that marks them as DMs. They are sent like any other posts, so there’s no encryption and instance admins could in theory read them in the database.
Not in theory, in practice, but this is not high need feature.
If you need to keep your sexting private, use another platform. If you have a exhibition fetish, go ahead.
As a coder I can say that e2e encryption is pain in the ass, key generation and exchanging, complex and annoying to do.
e2e encryption is pain in the ass, key generation and exchanging, complex and annoying to do
No, no it’s not.
Yes, it’s more complex than sending plaintext. But for starters it’d be extremely simple to generate a keypair for every user and publish the public key with their profile. When sending DMs you’d use this public key to encrypt the message.
As for storing the private key you could encrypt it with (a derivative of) the user’s password, and store it decrypted possibly just in the user’s browser.
This simple measure would prevent simple ways of reading the DMs, though obviously you still need to trust that your instance admins are actually serving you the code they claim they do. But it’d definitely prevent “accidental” misuse.
As for storing the private key you could encrypt it with (a derivative of) the user’s password
And now every time a user forgets their password and does password recovery, they lose all their DMs.
E2EE chat is a difficult problem.
That’s a feature not a bug!
Actually, users should not be required to trust the browser storage or in-app key generation, but be enabled to enter their own pgp key.I mean you could just store it encrypted in the database for the basics, and for advanced users allow them to back it up.
There are tons of ways to improve it, but there is definitely way more you can do without much inconvenience to the users. I doubt losing old DMs is a huge issue when you forget your password…
Fediverse adds level of complexity on it, like you mention.
Malicious Lemmy instance could man-in-the-middle by providing it’s public key in behalf of the user in other side. Normally this can be mitigated by CA, but CA doesn’t fit very well in decentralized system.
You could add AES with users own password, but problem is that same malicious instance could also steal users password.
IMHO false sense of privacy is worst than knowing that stuff is unsecure. Again in my opinion fediverse is comparable to yelling in town square.
- It is not unsafe.
- It is not 100% private. Admins can read your messages if they choose to investigate your messages.
- It will not get blasted out to the whole fediverse; just to the recipient you indicated. (Unless an admin from the previous point reads your message and publishes it publicly on the fediverse)
- You do not get to do anything naughty with it; expect to be caught if you break the rules.
It is not unsafe.
It is not 100% private. Admins can read your messages if they choose to investigate your messages.These points contradict one another.
How so? The message is safely delivered solely to the intended recipient, albeit in plain text (not private).
If there’s anywhere that the commonly used email analogy fits, I think it would be here
Safe and private are synonyms in this context.
I disagree.
Users likely trust their instances adminsUnless I run a Lemmy instance myself (which is possible), I have zero reason to trust an instance’s admins.
Even if my instance’s admin happens to be the founder of privacyguides.org, that doesn’t mean he will never read any “private” messages (or be forced by someone else to hand them over).
if you don’t trust the instance why would you use it? 🤨
Even if I did fully trust my instance, I also would have to trust any instance I message with.
I personally just use Lemmy for public comments.
What definition of safe are you using which makes a private messaging system without privacy safe? What would have to occur for it to become unsafe, if not being private does not make it unsafe in your eyes?
It does what it claims to do, it’s just that what it claims to do is clearly not complete privacy.
If something claims to be unsafe and delivers, that doesn’t make it any more safe.
No; they don’t. You just wanted to be a reply guy.
They absolutely do. A private messaging system which is not private is the definition of unsafe. Especially in the context of a post on !privacy@lemmy.ml
Anyone remember the days when they were called PMs? They are least used the word Private in the acronym.
What does DM even mean?
Direct Message. As opposed to Private Message.
Services at least acknowledge that the messages aren’t private, which is an improvement I guess.
Yes, and when we used the term “programs” instead of “apps” for everything.
Let’s go even farther back and say they were all “applications”
Not safe at all because it was not designed to be, it even tells you as much. Use them to exchange e2ee contacts and then use that.
@HiddenLayer5 @Kidplayer_666 U can use it to privately talk as much as ur comfortable with it going public.
I don’t know any of the technical details, but my understanding is they’re very unsafe. Our instance has a warning recommending that you take conversations involving sensitive information to Matrix
I dont think they’re supposed to be any safe at all. As part of our board culture we’re to expect that DMs offer no security, and to use other platforms if it’s a concern.