The idea is that you change or remove your username after someone else starts a conversation with you, so the username can no longer be used to subpoena your account details.
Put another way, signal is able to provide those 2 pieces of information to law enforcement based on a phone number. This helps you to prevent law enforcement having a phone number to ask signal to look up in the first place, assuming you change your username every time you hand it out.
They also hash the usernames that they store on your account which means law enforcement can’t ask what usernames are being used, only being able to ask for specific usernames which are currently in use.
I understand that right now LEA can serve up a subpoena and give Signal a username and get a phone number, but they can’t give them a phone number and get a username.
Is it also possible for Signal to keep track of past usernames/associated hashes for a particular phone number?
(For comparison, Signal could record IP addresses, but we trust they don’t due to unsealed cases. Could they keep a username history?)
Yes it entirely depends on whether they store previously used usernames along with the date range it was in use (to tell apart multiple people who used the same username at different times)
We’ll have to see if any unsealed cases in the future support that they don’t keep those records like how they don’t keep IP logs, but personally their track record is enough for me to have confidence in the feature, especially since my “threat model” is primarily opportunistic hackers or spearphishers at most, not police or state / nation state level actors.
They don’t track username history and don’t have a server side list of plaintext usernames, and others can’t find your phone number from the username alone. That makes it harder to confirm which account is yours.
Iirc from the last time this article or similar was posted, it’s about how warrants are issued. It’s the username versus phone number not username versus or and/or other data points. Anything more than that I am still unclear about.
And how does Usernames help here? Should be the same 2 data Points and not more?
The idea is that you change or remove your username after someone else starts a conversation with you, so the username can no longer be used to subpoena your account details.
Put another way, signal is able to provide those 2 pieces of information to law enforcement based on a phone number. This helps you to prevent law enforcement having a phone number to ask signal to look up in the first place, assuming you change your username every time you hand it out.
They also hash the usernames that they store on your account which means law enforcement can’t ask what usernames are being used, only being able to ask for specific usernames which are currently in use.
I understand that right now LEA can serve up a subpoena and give Signal a username and get a phone number, but they can’t give them a phone number and get a username.
Is it also possible for Signal to keep track of past usernames/associated hashes for a particular phone number?
(For comparison, Signal could record IP addresses, but we trust they don’t due to unsealed cases. Could they keep a username history?)
Yes it entirely depends on whether they store previously used usernames along with the date range it was in use (to tell apart multiple people who used the same username at different times)
We’ll have to see if any unsealed cases in the future support that they don’t keep those records like how they don’t keep IP logs, but personally their track record is enough for me to have confidence in the feature, especially since my “threat model” is primarily opportunistic hackers or spearphishers at most, not police or state / nation state level actors.
My phone number is registered to my phone carrier under my real name. My username is not. Unless I’ve misunderstood the question.
They don’t track username history and don’t have a server side list of plaintext usernames, and others can’t find your phone number from the username alone. That makes it harder to confirm which account is yours.
Iirc from the last time this article or similar was posted, it’s about how warrants are issued. It’s the username versus phone number not username versus or and/or other data points. Anything more than that I am still unclear about.