Spy agencies typically used small cells to avoid one compromise destroying the whole network. Organizers should copy this with big broad directions from the top, broad basic cooperation across the group but specifics as much as possible localized to small in-person cells of less than 10 people who all know and trust each other (e.g. not randoms who claim to be interested in the cause who contact you online but people known to others in the group so if some known reactionary who was posting Trump memes all last year wants to join you say no).
That and not using something that requires phone numbers which leads to network mapping and graphs which is what the NSA and CIA targeting have always cared about more than message content. Signal is compromised by US intelligence in the sense at least that it gives them maps of who is talking to who and associates them with real numbers they can connect to real identities thanks to phone company cooperation. Message content may be safe from broad collection but infiltration is still a problem. Not any good off the shelf solutions that don’t require hosting something or several somethings unfortunately that I’m aware of that aren’t equally suspicious as signal.
Bullshit. All they can see is if a phone number is registered to a Signal account. They get no visibility of social graphs. They publish the subpoena requests they are forced to comply with. Unless you have evidence proving otherwise, please retract your disinformation.
What exactly outbound signal message looks like to isp? if it’s a burst to central server with known ip, social metadata is absolutely trivial to extract, if it’s peer2peer (which seems exceedingly unlikely with phones constantly swapping ip), only then you have to do time correlation attacks and likely non-trivial to solve and easy to obfuscate. (p2p meshes also dodge direct inference of social connectivity, but i don’t think signal functions that way)
Not that they would compromise keyboard-encryption backdoor if it exists over something so trivial, but one shouldn’t just trust something on the say-so
The original claim was about phone numbers disclosing social graphs, but now we’re getting into network traffic analysis for a global passive adversary and a compromised device?
If you’re worried about traffic analysis use a mixnet like Nym. If you don’t trust your device, then get a device you trust.
No, it’s two claims: before signal servers and on your isp provider side sits nsa sniffing device (likelihood 99%), which trivially can reconstruct social graph without specifically designed obfuscations on server side (something like nym, exactly, but for signal servers themselves, with random delays and obfuscatory traffic).
Second claim is all messages are encrypted doesn’t exclude possibility of keyboard input->app internals middleman backdoor, likelihood of it existing unknown (hi, nsa), with pegasus infection 100% at least.
Basically, you are fucked with state adversary, and shouldn’t use phone for anything not serving to appear normal, and while signal can provide necessary tools to message each other, you should assume it to be transparent and appearing in some court if things go wrong. (But crucially transparent to very advanced adversary, not bumblefuck from local police, so it’s not a call to rely on messages which are so trivial to intercept with sim card duplication for 1k-5k bucks, and unencrypted for traffic interception)
Spy agencies typically used small cells to avoid one compromise destroying the whole network. Organizers should copy this with big broad directions from the top, broad basic cooperation across the group but specifics as much as possible localized to small in-person cells of less than 10 people who all know and trust each other (e.g. not randoms who claim to be interested in the cause who contact you online but people known to others in the group so if some known reactionary who was posting Trump memes all last year wants to join you say no).
That and not using something that requires phone numbers which leads to network mapping and graphs which is what the NSA and CIA targeting have always cared about more than message content. Signal is compromised by US intelligence in the sense at least that it gives them maps of who is talking to who and associates them with real numbers they can connect to real identities thanks to phone company cooperation. Message content may be safe from broad collection but infiltration is still a problem. Not any good off the shelf solutions that don’t require hosting something or several somethings unfortunately that I’m aware of that aren’t equally suspicious as signal.
Bullshit. All they can see is if a phone number is registered to a Signal account. They get no visibility of social graphs. They publish the subpoena requests they are forced to comply with. Unless you have evidence proving otherwise, please retract your disinformation.
https://signal.org/bigbrother/
What exactly outbound signal message looks like to isp? if it’s a burst to central server with known ip, social metadata is absolutely trivial to extract, if it’s peer2peer (which seems exceedingly unlikely with phones constantly swapping ip), only then you have to do time correlation attacks and likely non-trivial to solve and easy to obfuscate. (p2p meshes also dodge direct inference of social connectivity, but i don’t think signal functions that way)
Not that they would compromise keyboard-encryption backdoor if it exists over something so trivial, but one shouldn’t just trust something on the say-so
https://signal.org/blog/sealed-sender/
The original claim was about phone numbers disclosing social graphs, but now we’re getting into network traffic analysis for a global passive adversary and a compromised device?
If you’re worried about traffic analysis use a mixnet like Nym. If you don’t trust your device, then get a device you trust.
No, it’s two claims: before signal servers and on your isp provider side sits nsa sniffing device (likelihood 99%), which trivially can reconstruct social graph without specifically designed obfuscations on server side (something like nym, exactly, but for signal servers themselves, with random delays and obfuscatory traffic).
Second claim is all messages are encrypted doesn’t exclude possibility of keyboard input->app internals middleman backdoor, likelihood of it existing unknown (hi, nsa), with pegasus infection 100% at least.
Basically, you are fucked with state adversary, and shouldn’t use phone for anything not serving to appear normal, and while signal can provide necessary tools to message each other, you should assume it to be transparent and appearing in some court if things go wrong. (But crucially transparent to very advanced adversary, not bumblefuck from local police, so it’s not a call to rely on messages which are so trivial to intercept with sim card duplication for 1k-5k bucks, and unencrypted for traffic interception)