4 ms·
The great thing about no trust models is that even if they were comprimised it wouldnt matter
by jjcon 5y ago
The great thing about no trust models is that even if they were comprimised it wouldnt matter
- e12e 5y agoHardly. They get your password on login, and decrypt your mail. They don't (theoretically) even need a code change, just root/kernel level access/tracing to dump the password. Unless of course you were using PGP - but then you would be about as well off with Gmail? I mean, they provide a great service, and the stack is open - but I would hesitate to call it "secure" (that goes for ProtonMail too BTW). https://0xacab.org/liberate/trees https://0xacab.org/liberate/trees
- tgsovlerkhgsel 5y agoWhere is the no trust model there though? If riseup were a trap, they could do all the logging (in secret), find the activist, then do parallel construction for the actual evidence presented in court.
- basedrum 5y agoWhy would they voluntarily act as agents of the state to do that? They would need to be compelled legally to do so, and to do that in the US would require a legal order that has not yet succeeded (remember the Apple case around the terrorist in San Bernadino?). Before you say, "Well, it could be done in secret", please look up the Wiretap + technical assist laws and note the disclosure requirements.
- tgsovlerkhgsel 5y agoBecause if you don't trust them, it's plausible that some or all of the people running the infrastructure aren't actually privacy activists, but moles planted by the state, acting as agents of the state because they are employed by the state, paid by the state, and specifically tasked to pretend to be activists and build a trap. (There may be some solid evidence making this unlikely in reality, but that ultimately relies on some form of trust.)
- poetaster 5y agoTrees fails to load mobile?
- poetaster 5y agoI've never understood decrypting with login credentials. My gpg creds are distinct. My client challenges on each mail. Can be set to save for key per session, but don't.