I don’t think anything can be proven unless you have admin rights to the server at all times. signals are encrypted every time they are sent encrypted. can it be turned off with a flag? does it run in dev without it for troubleshooting and if so is it impossible to enable in prod.
exactly. if they had self hosted then it would be closer to equivalent to hilaries email but if it was using signal as written but then there is the foia issue which was still possible with hilarys email server, but not under a self hosted signal if not altered.
The server can’t decrypt it if it doesn’t have the keys to do so. It can be proven that private keys never leave the local device. It can also be proven that the proper public keys are being provided, and that the local device alerts on public key changes with a partner are announced.
Of course, nobody as part of the linked article did any of that verification, but still, a server doesn’t need to be trusted to be functional.
Doesn’t matter. Signal desktop app can sync messages and be installed on compromised computers. One of the guys in the chat was in Russia visiting Putin. It would be trivial to sync the account to the app installed on compromised machines and basically become an invisible backdoor into every secure communication for that user. I have no doubt one of the users in the chat is setup like this.
Oh, I’m not defending these dumb-asses doing illegal things to avoid systems setup to safeguard American and its people. They absolutely could have synced things to compromised devices. Just that Signal, themselves, couldn’t do that.
In the case of signal, it is provable that it cannot. They do not hold the keys to decrypt. The closest risk is the server injecting a new public key into the conversation, which the Signal app will warn about.
Signal does hold the public keys for every user. But having the public key doesn’t let you decrypt anything. You need the private key to decrypt data encrypted with the public key. So in a chat example, if you and I exchange public keys, I can encrypt the message using your public key, but only you can decrypt it, using your private key.
Signal does run the key exchange, which means they could hand a user the wrong public key, a public key which they have the private key for, instead of the other person’s. That is a threat model for this type of communications, however, signal users can see the key thumbprints of their fellow chat participants and verify them manually. And once a chat has begun, any changes to that key alerts all parties in the chat so they know a change has happened. The new key wont have access to any previous or pending messages, only new ones after the change took place.
I mean I still don’t see how it can be encrypted for a private key with the deryption at some point running through the server unless the members devices at some point communicate with each other without the server as an intermediary. Is that what happens at some point?
The way they explained things is how it works, do you not understand how public/private key pairs are used in encrypted communications?
Does this picture help?
I’m not trying to be an ass if my post comes across that way, I’m just unsure of your level of knowledge so I don’t want to offend by providing basic level info if it’s not wanted.
my experience with public/private keys is ssh where I have logins in both places and I login without keys and move it to the .ssh file. That is what gets me. how is the key moved over for communication and if the server moves it then it could theoretically hold onto it. So in the picture above it takes the public key in step 2 and sends it on in step 3 which is used to encrypt the message. I just don’t see how the public key in that case can encrypt the message but not decrypt it.
I don’t think anything can be proven unless you have admin rights to the server at all times. signals are encrypted every time they are sent encrypted. can it be turned off with a flag? does it run in dev without it for troubleshooting and if so is it impossible to enable in prod.
okay, so self host it if that is part of your concern/threat model. the Signal server code is open to the public, you can see and download it here.
exactly. if they had self hosted then it would be closer to equivalent to hilaries email but if it was using signal as written but then there is the foia issue which was still possible with hilarys email server, but not under a self hosted signal if not altered.
The server can’t decrypt it if it doesn’t have the keys to do so. It can be proven that private keys never leave the local device. It can also be proven that the proper public keys are being provided, and that the local device alerts on public key changes with a partner are announced.
Of course, nobody as part of the linked article did any of that verification, but still, a server doesn’t need to be trusted to be functional.
Doesn’t matter. Signal desktop app can sync messages and be installed on compromised computers. One of the guys in the chat was in Russia visiting Putin. It would be trivial to sync the account to the app installed on compromised machines and basically become an invisible backdoor into every secure communication for that user. I have no doubt one of the users in the chat is setup like this.
Oh, I’m not defending these dumb-asses doing illegal things to avoid systems setup to safeguard American and its people. They absolutely could have synced things to compromised devices. Just that Signal, themselves, couldn’t do that.
I think we are more talking about can the server decrypt the data. Not that the data is encrypted.
In the case of signal, it is provable that it cannot. They do not hold the keys to decrypt. The closest risk is the server injecting a new public key into the conversation, which the Signal app will warn about.
Yeah I just don’t get this. How does a person added to a chat get keys then?
Signal does hold the public keys for every user. But having the public key doesn’t let you decrypt anything. You need the private key to decrypt data encrypted with the public key. So in a chat example, if you and I exchange public keys, I can encrypt the message using your public key, but only you can decrypt it, using your private key.
Signal does run the key exchange, which means they could hand a user the wrong public key, a public key which they have the private key for, instead of the other person’s. That is a threat model for this type of communications, however, signal users can see the key thumbprints of their fellow chat participants and verify them manually. And once a chat has begun, any changes to that key alerts all parties in the chat so they know a change has happened. The new key wont have access to any previous or pending messages, only new ones after the change took place.
I mean I still don’t see how it can be encrypted for a private key with the deryption at some point running through the server unless the members devices at some point communicate with each other without the server as an intermediary. Is that what happens at some point?
The way they explained things is how it works, do you not understand how public/private key pairs are used in encrypted communications?
Does this picture help?
I’m not trying to be an ass if my post comes across that way, I’m just unsure of your level of knowledge so I don’t want to offend by providing basic level info if it’s not wanted.
my experience with public/private keys is ssh where I have logins in both places and I login without keys and move it to the .ssh file. That is what gets me. how is the key moved over for communication and if the server moves it then it could theoretically hold onto it. So in the picture above it takes the public key in step 2 and sends it on in step 3 which is used to encrypt the message. I just don’t see how the public key in that case can encrypt the message but not decrypt it.