4 ms·
this was rejected once. called big brother key in the lists if you wanna search for context and educate yourself better. argument was this workflow is already
by bdifbdifj 4y ago
this was rejected once. called big brother key in the lists if you wanna search for context and educate yourself better.
argument was this workflow is already possible. just encrypt an archive version as needed.
the original proposal asked for a always on key and was implemented with no user feedback. heh.
this uses a subkey on the original to key. less bad but still bad.
if more than one person must sign/decrypt, they should share the ownership of the private keys. you can already do these hacks where two private keys can unlock the secret to a third key etc. plenty of solutions.
this will just be used as a way to get a backdoor by offering less resistance on a corp environment. remeber, control comes not from forcing you to do something, but making you not care.
- upofadown 4y agoYou might be interested in this current discussion on the GnuPG mailing list: * https://lists.gnupg.org/pipermail/gnupg-users/2023-April/066503.html https://lists.gnupg.org/pipermail/gnupg-users/2023-April/066... >this was rejected once. called big brother key in the lists if you wanna search for context and educate yourself better. I found that the summary and the link at the end of the linked article was good enough. My understanding was that there was an actual bug in the implementation that caused a security weakness.