this post was submitted on 21 Jul 2025
68 points (100.0% liked)

Slop.

592 readers
484 users here now

For posting all the anonymous reactionary bullshit that you can't post anywhere else.

Rule 1: All posts must include links to the subject matter, and no identifying information should be redacted.

Rule 2: If your source is a reactionary website, please use archive.is instead of linking directly.

Rule 3: No sectarianism.

Rule 4: TERF/SWERFs Not Welcome

Rule 5: No bigotry of any kind, including ironic bigotry.

Rule 6: Do not post fellow hexbears.

Rule 7: Do not individually target other instances' admins or moderators.

Rule 8: Do not post public figures, these should be posted to c/El Chisme

founded 9 months ago
MODERATORS
 

https://archive.is/XtYXX

Denuvo won't be an issue forever. AI will be able to handle it easily within a decade. Hell, AI will be able to write emulators for the PS8 by the time the PS8 comes out.

Confidently incorrect redditor. I'm sure LLMs will be able to vibes code away a closed source system with next to zero documentation by design, that uses low-level machine code, several layers of encryption, and the world's foremost obfuscation made by the people who used to crack them. Not to mention that people like Empress have a vested financial interest in NOT disclosing their evasion methods.

you are viewing a single comment's thread
view the rest of the comments
[–] ZeroHora@lemmy.ml 4 points 3 weeks ago (2 children)

There's so many data about cracking Denuvo to train the IA in crack this shit, right?

[–] Taster_Of_Treats@hexbear.net 7 points 3 weeks ago* (last edited 3 weeks ago)

I honestly believe that to get the level of specialized machine code knowledge one would need for cracking Denuvo, you would have to devote a decade to studying it on your own. That or Empress would have to take on an apprentice. And she (or they if she is multiple people) is way too much of a narcissist and misanthrope to EVER do that.

To my (limited) understanding, the way that Denuvo obfuscates the machine code will poison the context window of an LLM. If the context window is full of nonsense that makes it seem like the DRM works one way but it works another way in practice, the LLM will weigh the incorrect and correct solutions equally and produce an incorrect solution. Obviously, I don't know how Denuvo works because nobody outside of Irdeto does at this point, but this is what Empress said and it's how peer software like Zimperium works.