this post was submitted on 02 Apr 2026
21 points (88.9% liked)
Technology
83452 readers
4767 users here now
This is a most excellent place for technology news and articles.
Our Rules
- Follow the lemmy.world rules.
- Only tech related news or articles.
- Be excellent to each other!
- Mod approved content bots can post up to 10 articles per day.
- Threads asking for personal tech support may be deleted.
- Politics threads may be removed.
- No memes allowed as posts, OK to post as comments.
- Only approved bots from the list below, this includes using AI responses and summaries. To ask if your bot can be added please contact a mod.
- Check for duplicates before posting, duplicates may be removed
- Accounts 7 days and younger will have their posts automatically removed.
Approved Bots
founded 2 years ago
MODERATORS
you are viewing a single comment's thread
view the rest of the comments
view the rest of the comments
Interesting, I wonder how much assembly these are trained on. I could see it working on the original source, with meaningful variable names, but when you have to figure out what r6 is used for in one code block based on where it's referenced, I don't see LLMs being particularly effective
It can likely untangle all the jumps an obfuscator makes with relative ease. After that it should be easier to decompile into something meaningful.
My point is that I'm not sure it can. Don't get me wrong, I'm at best a hobby grade reverse engineer (and genuinely pretty bad at it) but I don't think LLMs are well suited to that task. It does leave me kind of interested in testing it out. I see that someone has put ghidra mcp tolls together.