Where can I get the privacy nightmare AI application for Linux? I want it to take screenshots of my computer (in case I want to search it), never really use it, but store the data on my PC unencrypted, thus able to be hacked.
The security nightmare is reduced by a lot, thanks to Linux being a lot more safe system. Of course the occasional very old security issues get found, but those are only old if some swifty hacker found out and didn’t disclose it publicly, or had to wait for years to be solved.
I tried it to do some research recalls, like set it up for doing very specific things and I found it filled my hard drive pretty fast with screenshots. It’s probably a good idea if you can turn it on and off like this one and be careful, but it likely still needs polishing. That was when it was first out.
Python? This will require “specialized hardware” just due to the interpreter overhead taking continuous screenshots of everything you do and indexing/storing them. Why bother implementing something like this using an interpreted language??
When I first heard of the MS feature, my first thought was that there’s gotta be a more efficient way to do this than taking screen shots and analyzing the image. The window manager has all of that information plus more context (like knowing that these pixels are part of a non-standard window that uses transparency to act like a non-rectangular shape, while this thing that looks like a window is actually an image because the user was looking at someone else’s screenshot).
Even better would be integration with the applications themselves; they have even more contextual information than the window manager has.
If it required waiting for apps to provide this integration you would be waiting 10 years for it to mature. There would logically be some benefit to looking at the window before compositing + saving text from none visible windows and avoiding recapturing content which hadn’t changed.
Where can I get the privacy nightmare AI application for Linux? I want it to take screenshots of my computer (in case I want to search it), never really use it, but store the data on my PC unencrypted, thus able to be hacked.
Microsoft Linux.
I wonder if they’ll add this to WSL…
Someone made that, sort of. Unfortunately, the privacy nightmare is slightly reduced compared to the original one.
https://github.com/openrecall/openrecall
The security nightmare is reduced by a lot, thanks to Linux being a lot more safe system. Of course the occasional very old security issues get found, but those are only old if some swifty hacker found out and didn’t disclose it publicly, or had to wait for years to be solved.
I tried it to do some research recalls, like set it up for doing very specific things and I found it filled my hard drive pretty fast with screenshots. It’s probably a good idea if you can turn it on and off like this one and be careful, but it likely still needs polishing. That was when it was first out.
Python? This will require “specialized hardware” just due to the interpreter overhead taking continuous screenshots of everything you do and indexing/storing them. Why bother implementing something like this using an interpreted language??
Backups are important, especially passwords stored on images which is leagues more secure than plain text
If this was actually an open source app with strong privacy settings that you could opt into it might be useful.
When I first heard of the MS feature, my first thought was that there’s gotta be a more efficient way to do this than taking screen shots and analyzing the image. The window manager has all of that information plus more context (like knowing that these pixels are part of a non-standard window that uses transparency to act like a non-rectangular shape, while this thing that looks like a window is actually an image because the user was looking at someone else’s screenshot).
Even better would be integration with the applications themselves; they have even more contextual information than the window manager has.
If it required waiting for apps to provide this integration you would be waiting 10 years for it to mature. There would logically be some benefit to looking at the window before compositing + saving text from none visible windows and avoiding recapturing content which hadn’t changed.