I was taking a look at the Naomi Wu situation (A Chinese DIY tech youtuber who went missing after being watched by the government) and in one part they mentioned that she was concerned about her privacy, so started using Signal, but had a default chinese keyboard that had a keylogger and the police had looked into what she was talking on there.
I’m not sure if it was a mobile only thing, but it was mentioned that the keyboard app was used in like 70% por chinese smarthphones.
Now, I use AnySoftKey and refuse to use default keyboard apps, but how far can we reach on the keyboard security thing? Is typing on a computer or using a physical keyboard on a mobile device 100% safe? I think the keyboard issue is often overlooked and would like to know what recommendations your have? Or what should be known more?
On Wayland, they probably still can. Wayland’s core protocol doesn’t allow it, but extensions to enable things like global hotkeys can almost certainly be used for shenanigans.
Also, if the keylogger is running under your user account, it can insert crafted
.desktop
files wrapping around your apps,ptrace
your apps, you name it. Sandboxing as in Flatpak can stop this sort of thing, but if you run an app outside such a sandbox, and it’s malicious, game over.But does it work without prompting the user?
Also, I’m not too familiar with how it works, but afaik global hotkeys on KDE are implemented by the display server/compositor/whatever it’s called itself, and not sourced out to a different program.
Well, that’s an interesting point, I haven’t thought about that.
Right, but they’re configured by an unprivileged program: the settings app. Presumably, a keylogger can pretend to be the settings app.
Couldn’t the display server check if the app is actually the settings app by looking at it’s executable’s location? Not sure how reliable that is, but if it is, it could check if it is coming from somewhere in
/usr
that is also not writable by the current user.The display server has no way of verifying the process ID on the other end of the Unix-domain socket connection, and therefore cannot verify the executable image. It also cannot verify that the settings app hasn’t had any malicious code injected with
ptrace
,LD_PRELOAD
, or the like, since the injected code can remove any traces of that before connecting to the display server.