upvote
I think you're missing group 3, which is the "assume bad faith and/or bad outcome from any possible change to anything, and especially changes to open source projects that can be interpreted as exogenous, e.g. from an acquirer, and especially if that acquirer is a gigantic commercial entity."

There are not nearly enough disenfranchised Bun-on-Zig contributors to make a dent in this conversation. There are lots of Group 3s in every similar convo, for any combination of technology, project, and acquirer you can name.

reply
I think most of the naysayers are in group 3 because there’s a lot of anger but never a single link to a blog post or analysis done by anyone to demonstrate a regression. not one. There are some that analyzed the quality of the initial rust code and use of unsafe but Claude has been chewing through those as I understand.
reply
I thought that too but it looks like the usage of unsafe hasn't been trending down over time: https://news.ycombinator.com/item?id=48966569#48967630
reply
https://github.com/oven-sh/bun/issues/30719#issuecomment-446...

Personally I chose not to disclose my findings in public to avoid this kind of outcome. Having a good laugh with others is not as important as keeping the result of this experiment undisturbed.

This also eliminates the chance of another round of marketing of LLM ability based on actual human contributions and I suggest everyone who is able to analyze the code to do the same - so that they cannot misattribute our ability to LLMs.

reply
It was a PR issue more than anything else. It is also a proxy for Pro AI and Anti AI sentiment.

And for most part I don't think it was communicated well apart from the last blog post.

reply
What about Group x.

You just AI generated 1 millions lines of code claiming it's for safety. Who exactly is to make any kind of security guarantees about this?

reply
What was your safety guarantee pre-rewrite? If CVE scanners and aggregators is your answer, then well, this applies to post-rewrite as well.
reply
deleted
reply
deleted
reply
Don't forget the group asking really dumb pointless questions like "who legally owns the rewritten code"
reply
Afaik, under current US law (Thaler v. Perlmutter 2025^), "the Copyright Act of 1976 requires all eligible work to be authored in the first instance by a human being".

With regards to AI-generated code, this means the copyrightability (at all, not with regard to who owns it) turns on whether or not a human was substantially involved in its creation.

Existing decisions require evidence of pretty heavy and continued human guidance to qualify.+

To wit, autonomous agent created code (prompt -> machine churns for hours -> output) is explicitly not eligible for copyright.

Functionally, this is a double-edged sword.

On the one hand, it means anything coded with autonomous agents by Meta, Google, et al. can be legally reused if it leaks (because no one could hold copyright on it).

On the other hand, it leaves copy-left open source licenses in a weird place. If you convert open source (even MIT/BSD-style) code into something else with an autonomous coding agent... the result has no copyright (nor can ever in the US).

In this instance, Bun was MIT-licensed, no? Then it was shoved mostly-autonomously through an LLM for the port.

Now Bun-Rust is technically still MIT licensed, but if push came to shove it seems like US law's current position is that Bun-Rust would now have no copyright license (because the manner in which it was developed renders it ineligible for copyright).

That's on the copyright side.

On the infringing usage side (i.e. whether you were entitled to shove a copyrighted work into a coding agent to produce something)... that's still TBD.

^ https://media.cadc.uscourts.gov/opinions/docs/2025/03/23-523...

+ https://garrettham.com/ai-generated-works-copyright/

reply
Couldn't it be argued that Bun-Rust is still MIT licensed, because the old implementation was "in the first instance" authored by a human? All the machine did was translate it.

I think it's the "infringing usage" question that is more interesting. If the LLM trained on GPL-derived code, what does that mean for the end result?

reply
"If the LLM trained on GPL-derived code, what does that mean for the end result?"

Either it means nothing at all, or it means that a substantial portion of the code produced at some of the world's most valuable companies over the last two and a half years is GPL.

reply
Yeah you could argue that, but it's a messy argument. I'm wiling to imagine that they hadn't used AI to do the rewrite to Rust. It wouldn't have been all that hard to do with plain old syntax tree transformation (my expertise). The trickier question for me is the ownership at the boundary where prompts were turned into Zig code. That is not a process that could have been done with syntax tree transformation: creative freedom was given to the model as to how to do the work at that stage, and so that is the stage at which I'm least sure they could legally defend their claim to ownership.
reply
I feel like that's a lot of LLM use in a nutshell: sure, there's a more surgical scalpel available, but isn't it so much easier just to fling prompts at a magic box?
reply