LLMs can read and write executables byte by byte from a hex dump
91 points by fortran77 4 days ago | 70 comments

pier25 4 hours ago
I've suspected for years that AI companies have been using commercial software to train their models including games, audio plugings, DAWs, etc. It would not surprise me one bit if we find out they used torrented pirated apps like they did with books and music.

There's certainly eg open source audio dev material out there but not nearly as much as web dev. How else do you explain all the avalanche of vibe coded audio plugins and DAWs appearing this year?

reply
dachris 2 hours ago
And (from a training material point of view) it makes perfect sense.

I would assume that to a LLM, a binary executable is not much worse than actual source code. And it's much denser, just missing the semantics of human-written identifiers. It's the code's pure logic, condensed into machine language.

And there's tons of it, many vendors supply it "freely" (you don't need license keys to train on a binary) It's often battle-tested code that you wouldn't get hold of otherwise, and compared to ingested source code its unlikely that Codex regurgitates a recognizable part of a Windows 2000 DLL into your C++ source code.

reply
mycall 37 minutes ago
Being deterministic, a binary executable is just a compressed version of source code.
reply
azornathogron 9 minutes ago
Compressed doesn't sound quite right. Source code contains abstractions which are often then expanded out into many variants in an executable (most obviously by function inlining, but at a smaller scope also by things like loop unrolling etc)
reply
shermantanktop 14 minutes ago
Is plate of spaghetti a compressed version of a recipe?

Executables are often larger than their source code.

A toolchain that creates repeatable binary builds is difficult to create. And a nonrepeatable build is potentially nondeterministic, in the sense that uncontrolled or hidden inputs impact the output.

reply
xnx 3 hours ago
My hunch is that one of the handicaps Google is dealing with (vs. OpenAI and Anthropics) is that Google is not training on pirated media. They are using YouTube, Google book project scans, and purchased copies of media.
reply
walthamstow 2 hours ago
Perhaps they are not actively collecting it like Ant/OAI but YouTube and Google Books are full of user-uploaded pirated media.
reply
kakacik 40 minutes ago
OP is talking about binaries
reply
mixtureoftakes 3 hours ago
There is zero chance models are not trained on pirated media and games, especially most popular ones.

Makes complete sense to study things people actually want to use and consume

reply
basilgohar 3 hours ago
I agree 100% with sentiment. It is inconceivable to me that models were not trained, either directly, or indirectly through other aggregates, on code and binaries and, of course, media, fetched from torrent or other filesharing sites.
reply
Keyframe 3 hours ago
wouldn't that information leak out or somehow found out? too many people already involved. With some, like Seedance it's obvious of course.
reply
docshock 2 hours ago
I thought the fact they fed copyrighted material into the models has been established.
reply
singpolyma3 3 hours ago
Not saying you're wrong, but LLMs can work in areas they've never been exposed to in languages they've never been exposed to. Training is not needed for something to be possible
reply
WarmWash 3 hours ago
They also use synthetic generated data to train on too

If it works and it's verified in nine different ways, throw it in the training pile, and then automate the process.

reply
devmor 3 hours ago
There is absolutely no way to prove that this claim is true because no model advanced enough to do so at a useful level has any fine grained report of its training data that could be used to verify this.
reply
djleni 2 hours ago
This is trivially true. Why can they work with a tool or API I’ve written or something internal to a company? They definitely have not trained on my local machine from 2020.
reply
docshock 2 hours ago
I'm pretty sure companies are not throwing away their training data each time they build a model.
reply
petu 3 hours ago
> How else do you explain all the avalanche of vibe coded audio plugins and DAWs

By them being nothing special?

> appearing this year?

Vibe coding in general became viable last year, Opus 4.5 is Nov 24, 2025.

reply
SyneRyder 2 hours ago
Also the 1 Million token context window helped make it more viable.

I've made a bunch of unreleased vibe-coded VST plugins to replace the hundreds that I've bought over the years - and many of the barriers to development weren't the model. Just having belief that the model was capable of the work, a bit like the 4-minute mile. Also knowing to persevere and iterate with the model, not giving up because the first UI is inevitably so amateurish.

There's also a flywheel happening. If you make permanent notes on the process, your rules for UI design, a library of re-usable components and algorithms, tools so the model can judge its own work... the whole process for the next VST just gets much faster & much easier. Same goes for any software you make with AI too.

reply
whateveracct 2 hours ago
so weird how AI boosters act like it hasn't been less than a year, isn't it?
reply
bitwize 2 hours ago
It hasn't been. Since Opus 4.5 "letting Dario take the wheel" became useful for the vast majority of programmers.

As of this fall, you are being professionally irresponsible and SHOULD NOT be working in the field if AI is not integral to your SDLC. At a minimum it should be checking your code for bugs and vulnerabilities.

reply
fwlr 57 minutes ago
There are legal issues, ethical issues, moral issues, aesthetic issues, economic issues, financial issues, availability issues, longevity issues, consistency issues, reproducibility issues, sovereignty issues, regulatory issues, licensing issues, terms of service issues, and data retention issues that need to be addressed before AI can even be permitted in my software development life cycle - much less made mandatory on pain of unemployment.
reply
bitwize 24 minutes ago
What about security issues? What about the fact that if AI isn't hardening your code under your control, it's hacking it under an adversary's?
reply
whateveracct 36 minutes ago
this is what i'm talking about
reply
rhdunn 3 hours ago
It should be theoretically possible to use C/etc. source code, assembly output, and compiled output in the training set with equivalence mapping between them. That would make it possible for the LLM to learn disassembly and decompiling to some degree.

You don't need software for that aside from the compiler/linker tools (which are free for C/C++ compilers, rust, etc.). You can then permutate/generate the source code using tools/techniques like fuzzing.

There's also things like Intel's instruction set documentation.

reply
cedws 3 hours ago
An LLM trained on decompilation would be a game changer, but there's no way they'd allow us to use it.
reply
ape4 2 hours ago
Maybe you could use a regular decompiler. It makes bad variable and function names. Then get an existing LLM to correct the names.
reply
dreamcompiler 2 hours ago
It's been a thing for a couple of years now e.g.

https://arxiv.org/html/2403.05286v2

reply
sorokod 51 minutes ago
torrented pirated apps may contain all sorts of nasties, it would be amusing if the nastiness gets embedded in the LLMs
reply
saagarjha 2 hours ago
AIs are quite good at using existing binary analysis tools; they usually do not just operate on binaries in “prompt space” like this
reply
riggsdk 3 hours ago
In that case it would be movie-plot worthy when the LLM was trained on all those embedded viruses and malware. "Believing" it is intentional and then start embedding it into all software it creates for us.
reply
goodmythical 10 minutes ago
report shows AI botnet is using your fridge as distributed command and control server

"Hello, we're here to see whether or not your fridge is runni-Hey, no not like that, this is serious!"

reply
jrmg 10 minutes ago
This has been true for some time now. I can't remember details, but about a year ago, I was debugging some boot issues on an AVR microcontroller, and just pasting the HEX of the bootloader (that I didn't have access to the source of) into Claude's web interface got me the answers I needed.
reply
MomsAVoxell 3 hours ago
Back in the day, thats how we got executables from one system to another, over ASCII-tuned serial lines. I had a colleague in the 80’s who could muscle-memory a “hx.c” which would produce the requisite hex listing given an input, and if that input was a hex listing, reproduce the binary. It was something like 12 lines of C code and it saved our asses so many times, he eventually made a DOS compatible equivalent that could be piped into DEBUG.com to produce the same results.
reply
euroderf 2 hours ago
IIRC there was a bastard mini version of uudecode that was written so that all the opcodes were printable ascii, so you could save the text block, mark it as executable, and off you went... to get a uuencode of the full uudecode binary, decode it with your bastard mini uudecode, and then you were all set.
reply
plastic3169 18 minutes ago
It feels like image models which create images and video by bitbanging pure pixels. This is almost never the way people would approach the task of creating images barring pixel graphics. It also means that using image models is inconvinient. I always likened it to having programming AI that would go from prompt to binary. I thought we don’t program like that because programmers understand the problem area better and like to go through source code.
reply
goodmythical 5 minutes ago
000000?

Nope.

000001?

Nope.

...

000000 00001 00010 00110 00000 100000?

Hey, we've got addition!

reply
bita_nidir 3 hours ago
I don't think it'll make sense for an LLM to program directly on the cpu opcode level due to the overhead.

But I hope you all realize that the programming language of the future is going to be for LLMs. It'll tightly optimized for LLMs and we'll get nice tools to understand them. They'll probably be great, with loads of fancy tooling.

reply
amelius 2 hours ago
The programming language of the future will be a purely functional language, where the LLM acts as the compiler using correctness preserving transforms.

Any other way of coding will be considered ridiculous (except maybe if you do it for a hobby).

Better start learning Haskell, folks.

reply
WarmWash 2 hours ago
In its purest form there would be no programming at all, you would just be interacting directly with the AI model. The interface would be like an amorphous digital goo that can instantly shape-shift into any state.
reply
massysett 41 minutes ago
A learned comment in this forum stuck with me: suppose diff has not been written. You can use an AI to write diff. Or you can use AI to do a diff. The latter might be easier, but now you have to pay every time you diff files.

But as AI proliferates and gets cheaper, why not pay to diff files. It will be sufficiently cheap.

reply
goodmythical 2 minutes ago
I will never touch a computer again the moment I must pay a rental fee for diffing.
reply
amelius 31 minutes ago
At that point, just ask AI to write the Holodeck and we won't ever need to use computers anymore.
reply
bonesss 2 hours ago
OCaml, F#, or even Rust have functional cores that verify correctness. They’re wire token efficient, too.

Sufficiently massaged OCaml/F# approaches straightforward readability in plain English.

And, for people who have been doing it for a minute, the other ways of programming do seem a little ridiculous ;)

reply
sevenzero 2 hours ago
Yea lets just trash all these legacy codebases the world depends on, I'm sure this will work out
reply
raincole 2 hours ago
You're being sarcastic but it's likely necessary. There will be so many vulnerabilities found in legacy codebases and not enough people who understand them to patch the holes. We'll have to rely on LLM to update them and the once well-understood codebase would be more and more like a vibe-coded one.
reply
fwlr 47 minutes ago
Some people, when confronted with a problem, think “I know, I’ll use AI.” Now they have many problems. They think “I know, I’ll use many AIs.”
reply
amelius 2 hours ago
Legacy codebases are toast. This is where the software world is heading:

https://www.youtube.com/watch?v=Bo-vESS4jt0

https://www.youtube.com/watch?v=UCL4K36SHp0

reply
behringer 2 hours ago
It's like watching a boat pull into port. You can see the boat on the horizon, but people are saying the boat won't be arriving any time soon.

The boat's here. Nobody's going to be programming the old fashioned way anymore (except as a hobby or a specialist). You should probably get ready for the boat.

reply
duskdozer 59 minutes ago
They have to go. You must keep feeble human hands off the code as much as possible. Ideally, all coding will be done in a safe and secure proprietary binary format owned by Anthropic, obfuscated and rich with LLM-poison booby traps to ensure that nobody can attempt to modify it using inefficient meatcoding or dangerous and unsafe bootleg Chinese LLMs. Only then can we achieve maximally aligned and efficient GSI and infinite prosperity for all.
reply
bitwize 2 hours ago
Claude can rewrite those codebases in Rust in an eyeblink. Let them all burn.
reply
andai 2 hours ago
What overhead? More tokens for same result?

I haven't looked into it properly but, it looks like it might actually be the same amount of tokens? (Excepting stdlib boilerplate etc.)

reply
haolez 3 hours ago
It depends. If you design and train a tokenizer for that, I don't see why it wouldn't work, and I'm not even an "AI psychosis" victim :D
reply
dezgeg 2 hours ago
Emitting a executable binary (eg. ELF) will never make sense over using assembly language. Adding/removing some code to the middle of a large program shifts tons of addresses like jump targets and variables. It makes no sense to have LLM spend tokens on making edits to (potentially) every single call instruction in an existing binary, instead of letting assembler do it.
reply
rossy 2 hours ago
Like the weight space stuff, I think there's a small misdirection in how this is presented: Phrases like "write directly in machine code" seem to imply that ordinary LLMs intuitively understand machine code as just another language, like English or JavaScript, that they read from context and write directly as tokens, and that this intuitive understanding could mean increased efficiency or capabilities if you ask the LLM to just use the bytes rather than asking it to use the tools that human reverse engineers use.

If that's what you assumed, the DeepSeek reasoning trace shatters that illusion. It's just disassembling by hand like what a human reverse engineer would do if you (sadistically?) denied them a disassembler. Yeah, the human reverse engineer would need an excellent memory of the x86_64 instruction reference, but we already know LLMs can recall a breadth of information. What we don't know, that this article also provides no evidence for, is if there will be any benefit in using the LLM for this over objdump.

The article seems to intentionally conflate disassembly (computationally cheap) with decompilation (can be computationally expensive) to imply that there's some benefit to asking the LLM to work with machine code (seems impressive to the RE layman) rather than working in assembly (something you probably already assumed LLMs can do,) but the DeepSeek trace has a disassembly in it, and it's hard to imagine that inference will ever be cheaper or more reliable per-token than objdump is at producing the same output.

reply
mfuzzey 3 hours ago
This can be useful in the reverse engineer / malware analysis case where there is existing machine code and no source but I don't think that AI "writing machine code directly rather than high level languages" makes sense. Code generation by LLMs is a non deterministic process so it makes far more sense to generate higher level compilable source code once and then compile it with a deterministic compiler for each target rather than generate machine code non deterministically several times seperately for each target.
reply
saagarjha 2 hours ago
We also have more or less deterministic tools to lift code as well, which are usually more efficient to use.
reply
JKCalhoun 3 hours ago
An uninteresting case I am sure, but I have been leaning on LLMs to disassemble 6502 hex dumps and it's no doubt easy-lifting for them.

And the proof of course is that I can reassemble/link, with "modern" tools, their presented assembly and it combines right back into a hex dump mirroring the original.

Anyway, it's been a joy to take the long-out-of-print book, "The First Book of KIM", and re-typeset it (re-assemble it?) for modern hobbyists like myself. LLMs have allowed me to clean up the poor OCR-extracted text from the original. Claude has not only done that, but found missing labels and "equates" in the assembly, as legitimately printed in the book—not OCR errors.

Heck, Claude latched on to a bug in the classic "WUMPUS" game that apparently the author was not even aware of. (But I left that bug in for hobbyists to find—or exploit.)

reply
Neywiny 3 days ago
Wow. It can read bytes from a byte output? That's incredible
reply
plqbfbv 2 hours ago
The incredible part is that an LLM that wasn't explicitly trained for this, can analyze bytes as if they were natural language, and surgically modify and/or insert bytes directly without going through intermediate steps (source -> ASM -> ...).

This process would normally require a human to lift bytes -> ASM, ideally ASM -> obfuscated source, making sense of it, and doing the chain backwards, and it requires coordinating a lot of knowledge. Which of course is no problem for an LLM.

reply
saagarjha 2 hours ago
People can do this in some limited circumstances, LLMs are just better at it
reply
nubinetwork 2 hours ago
A hex dump is just extra bytes of ascii filling up context... having to deal with "20 e3 8f" is 8 characters of text, compared to 3 bytes. LLMs can already parse binary files without the text overhead.
reply
andai 2 hours ago
Couldn't they do this before?

They don't list any model size or version that failed, right?

reply
flohofwoe 2 hours ago
Hmm, is this new or surprising? When testing my home computer emulator code for correctness, Opus had no trouble taking C64 .prg files apart and analyse them the same way it would analyse .asm source files.

PS: and it also has no problem writing a simple 'bouncing sprite' C64 demo without going through assembly and wrap that in a .prg file (of course this is basically the 'hello world' of C64 demos and could be as-is in the training set).

Let's see if the data URL survives HN:

https://floooh.github.io/tiny8bit/c64.html?file=data%3Aappli...

reply
layer8 3 hours ago
I wonder if prompt injection via an executable is possible. Especially in the context of having AI perform malware analysis.
reply
saagarjha 2 hours ago
Yes but models are fairly resistant to simple prompt injection these days. Having them hit guardrails is a more common strategy
reply
tliltocatl 2 hours ago
So can I. CD 20
reply
dreamcompiler 2 hours ago
Since disassembly is a solved problem and it's deterministic with a 1:1 solution, why throw a nondeterministic LLM at it? Why not give the disassembly output to the LLM and let it do the decompilation, which is not 1:1?

Pet peeve: Using AI to solve problems that existing algorithms can dependably solve with 10^-6 the CPU resources.

reply
WhereIsTheTruth 2 hours ago
The point is you no longer need expertise as a prerequisite
reply
secondcoming 4 hours ago
I recently watched a video of a guy trying to unbrick his Harley Davidson Livewire electric motorbike. Harley Davidson were unable to do it. Perplexity created an app to sniff the canbus and talk to the battery and other components on the bike. Soon it'll be possible to write your own firmware for things like this.

[0] is the episode I'm referring to, but the whole series of videos about his woes with this bike are quite interesting.

[0] https://www.youtube.com/watch?v=Z4Ty0oqA_Bk

reply