this post was submitted on 18 Sep 2023
20 points (100.0% liked)

technology

23182 readers
498 users here now

On the road to fully automated luxury gay space communism.

Spreading Linux propaganda since 2020

Rules:

founded 4 years ago
MODERATORS
 

they fixed it :(

top 12 comments
sorted by: hot top controversial new old
[–] stardust@hexbear.net 12 points 1 year ago (1 children)

Activating Windows without a key is already super fucking easy

[–] Mehrunes_Laser@hexbear.net 1 points 1 year ago

Yeah I used a powershell script to activate mine.

[–] FunkyStuff@hexbear.net 9 points 1 year ago (1 children)

data-laughing I'm so sad that they fixed it. It can still generate Borderlands 2 Shift codes, though.

[–] roux@lemmygrad.ml 8 points 1 year ago (1 children)

Say what??? Like do the shift codes work?

[–] FunkyStuff@hexbear.net 6 points 1 year ago (1 children)

Oh I don't know but it did generate them. If you try to get it to do Windows keys it just tells you it can't.

[–] DayOfDoom@hexbear.net 2 points 1 year ago (1 children)

Is it generating or just reposting codes already on the internet somewhere?

[–] FunkyStuff@hexbear.net 2 points 1 year ago

Generating. In theory, it has some way to find patterns in how legitimate Windows keys are generated, and it's reproducing those patterns even without knowing what the algorithm that produced them is actually doing. That means it takes a bunch of tries before you get one right because they are only educated guesses. Pretty cool that it can do that though and I wonder what else it can try to do.

[–] kristina@hexbear.net 5 points 1 year ago (1 children)

is it confirmed that they fixed this lmao

[–] FuckyWucky@hexbear.net 4 points 1 year ago (1 children)

yea i tried asking chatgpt sadness

[–] kristina@hexbear.net 3 points 1 year ago (2 children)

how many iterations did you try, or did it not work at all

[–] FuckyWucky@hexbear.net 4 points 1 year ago

it doesnt reply at all. says it wont do anything illegal.

[–] FunkyStuff@hexbear.net 4 points 1 year ago

Can confirm they patched it and I tried a few different methods. That being said, with the way all this stuff works I can see 2 cases:

  1. They just hard coded it to shut down whenever the user prompts it with some combination of "Windows" and "Keys", in which case Chat GPT can still be exploited in similar ways for a ton of other fun piracy uses.
  2. They made it "intelligently" detect when the user was trying to trick it, which is what it (nominally) has always tried to do, so there's still a billion ways to get it to give away sensitive info because AI don't real.