i think this sort of scenario is where an ai code comb through is going to add value - get an ai to guess the likely name of variables/functions based on usage and context etc. It'd make a great starting point for manually labelling each variable correctly.

I'm using Ghidra with Claude-Code to reverse engineer an old transportation-sim game, and Claude is very good at figuring out what a function does and what it and its variables should be named.

I’d argue using AI to provide that input is no longer a creative work and puts the output in the realm of being not transformative. Those are two of the main reasons these projects are not violating copyright and all of the legal risk that comes with decompiling.

It's currently legally grey as to what ai output constitutes.

I'd argue that copyright should not be affected by the tools being used. If the project would've been compliant with copyright (ala, fair use) had it used human hands, then using an ai tool makes zero difference.

USCO disagrees. Output of an LLM is not considered a new, original work. https://www.insidetechlaw.com/blog/2024/05/generative-ai-how...

Usually it means it has bit-perfect recompilation, which can take a lot of work.

This is very wrong. Ghidra might decompile to some C code, but this is a completely different, and very high bar. This is 100% matching C code that compiles to the exact binary the game shipped with. This means it's able to be modified, tweaked and upgraded. Ghidra is helpful to do some research, but it won't compile.

That's not what the parent comment said: "100% decompiled to C, but not fully labelled yet".

Getting the exact same binary is definitely something else. Figuring out the exact compiler version and options, libraries, etc. is itself nontrivial.