Is that not the LLM guy? [LWN.net]

Is that not the LLM guy?

Posted May 8, 2026 18:41 UTC (Fri) by rahulsundaram (subscriber, #21946)
In reply to: Is that not the LLM guy? by mirabilos
Parent article: killswitch for short-term emergency vulnerability mitigation

Other people clearly don't share your views on LLM. Not confusing really especially since https://docs.kernel.org/process/coding-assistants.html clearly documents the views of the project itself

to post comments

Is that not the LLM guy?

Posted May 8, 2026 19:37 UTC (Fri) by IanKelling (subscriber, #89418) [Link] (5 responses)

(mostly sarcasm) Confusion abounds. Is there a correlation between the LLM assistance in this patch and the KILL terminology in this patch? Would a less amoral / llm unassisted contributor have named it differently?

Is that not the LLM guy?

Posted May 11, 2026 1:03 UTC (Mon) by cypherpunks2 (guest, #152408) [Link] (4 responses)

Is there actually a problem with the name killswitch? I didn't see any criticism of that on LKML.

Is that not the LLM guy?

Posted May 13, 2026 0:25 UTC (Wed) by IanKelling (subscriber, #89418) [Link] (3 responses)

Use your own mind and emotions! To have an LLM listed as co-author for a function called killX, where X is just about anything, that seems pretty ripe for a joke at least. I said "mostly", well, because humans doing killing of other living beings is an ordinary fact of life, automated killing using llms seems likely be an inhumane horror worthy of global attention, so, LLM listed as coauthor/"Assisted-by" on this seems maybe just a little thoughtless or distasteful in a strange way, which also seems to be one of the biggest problems with LLM generated content in general.

Is that not the LLM guy?

Posted May 13, 2026 0:48 UTC (Wed) by sashal (✭ supporter ✭, #81842) [Link] (1 responses)

Some folks on this thread got it, but in case you missed it, it was a reference to https://en.wikipedia.org/wiki/Killswitch_Engage

Is that not the LLM guy?

Posted May 13, 2026 17:21 UTC (Wed) by IanKelling (subscriber, #89418) [Link]

I missed it. Thanks for clarifying and please excuse my wondering from the peanut gallery, I'm happy to learn it was just my imagination.

Is that not the LLM guy?

Posted May 13, 2026 5:17 UTC (Wed) by zdzichu (subscriber, #17118) [Link]

Is Poe Law active here?

Is that not the LLM guy?

Posted May 8, 2026 21:06 UTC (Fri) by mirabilos (subscriber, #84359) [Link] (12 responses)

Oh, but that document says…

> * All code must be compatible with GPL-2.0-only

… which clearly excludes LLM-regurgitated slop.

After all, it is a plagiate of other works under conflicting licences and, in ipso, unlicenced.

Is that not the LLM guy?

Posted May 8, 2026 22:28 UTC (Fri) by intelfx (subscriber, #130118) [Link] (9 responses)

>> * All code must be compatible with GPL-2.0-only
> … which clearly excludes LLM-regurgitated slop.
> After all, it is a plagiate of other works under conflicting licences and, in ipso, unlicenced.

It must be noted that this (both the premise and the conclusion you made from this premise) is your opinion only, which is not necessarily shared neither by the majority of the community, nor by the law and/or legal precedent so far.

Is that not the LLM guy?

Posted May 8, 2026 23:15 UTC (Fri) by mirabilos (subscriber, #84359) [Link] (8 responses)

At least the second half is confirmed by the copyright office of the USA.

Is that not the LLM guy?

Posted May 9, 2026 0:56 UTC (Sat) by intelfx (subscriber, #130118) [Link]

> At least the second half is confirmed by the copyright office of the USA.

Which one specifically? And [citation needed].

Is that not the LLM guy?

Posted May 11, 2026 20:13 UTC (Mon) by JoeBuck (subscriber, #2330) [Link] (6 responses)

Are you sure? The US Copyright Offfice has said that AI generated works aren't protected by copyright, which basically means public domain. So if a patch is partly AI-generated and partly human-written, and the human contributor wants to use GPL 2.0, then the combined work is GPL plus public domain code. So, legal to distribute under the GPL.

Of course, IANAL.

Is that not the LLM guy?

Posted May 11, 2026 20:28 UTC (Mon) by farnz (subscriber, #17727) [Link] (1 responses)

Have you a link for this?

The last I saw, the US Copyright Office had said that the AI cannot be a copyright holder on an AI generated work, and thus that where the attempt at registering a work explicitly states that the AI is the sole author, there is no copyright.

However, this does not mean that AI generated works aren't protected by copyright; where the AI is used as a tool by a human, there's room for the human input to be enough to reach the bar needed to make the work protected by copyright.

It's essentially, in law, the same position as Emacs; if you claim that the work is purely produced by Emacs, then it can't be protected by copyright (since Emacs is a tool, not a human), but if you claim that the work is produced with Emacs driven by a human being, the resulting work can be protected by copyright. Substitute "AI" for "Emacs", and you get the answer the US Copyright Office came to, last time I looked.

Is that not the LLM guy?

Posted May 11, 2026 20:55 UTC (Mon) by mirabilos (subscriber, #84359) [Link]

> The last I saw, the US Copyright Office had said that the AI cannot be a
> copyright holder on an AI generated work

That, yes (that much is obvious, it has to be human creativity), but they did NOT rule out that copyright from authors of “training data” applies.

Is that not the LLM guy?

Posted May 11, 2026 20:30 UTC (Mon) by Cyberax (✭ supporter ✭, #52523) [Link] (3 responses)

It's likely that we'll see the protection applied to the _prompt_.

So if you ask the LLM to "fix this issue", then the result is not going to be copyrightable. But if you provide a detailed step-by-step spec for the fix, the result might be protected.

Is that not the LLM guy?

Posted May 12, 2026 10:17 UTC (Tue) by stijn (subscriber, #570) [Link] (2 responses)

Prompts do not deterministically lead to the same result every time, so the detailed step-by-step spec is not a spec; I have doubts that the code resulted from the prompt is thus copyrightable. The rules in this new world are very unclear to me.

Is that not the LLM guy?

Posted May 12, 2026 19:25 UTC (Tue) by Cyberax (✭ supporter ✭, #52523) [Link]

Sure. And in photography you'll never get two identical pictures. Yet you can still copyright specific arrangements of items in a photo.

This is going to be a case-by-case issue.

Is that not the LLM guy?

Posted May 12, 2026 21:27 UTC (Tue) by rgmoore (✭ supporter ✭, #75) [Link]

The question is whether the LLM is the thing doing the work or if it's a tool being used by a human. Deterministic results aren't necessary to show human creativity was involved. Plenty of artists use techniques that incorporate some degree of randomness, like Jackson Pollack's splatter paintings or works including found objects. As long as the artist is making substantial creative choices, the work still qualifies for copyright. The same thing should be true of using an LLM as a programming tool. If the programmer is providing the ideas and the LLM is only handling implementation details, there would probably be enough creativity to qualify.

Is that not the LLM guy?

Posted May 9, 2026 13:19 UTC (Sat) by jejb (subscriber, #6654) [Link]

> > * All code must be compatible with GPL-2.0-only
>
> … which clearly excludes LLM-regurgitated slop.
>
> After all, it is a plagiate of other works under conflicting licences and, in ipso, unlicenced.

Well, that's a whole load of confused statements. As long as a model output isn't derivative of its training data (as court cases have so far mostly confirmed) the US copyright office maintains that machine authored code is uncopyrightable. So that means public domain and combinable with any licence including GPL-2.0.

Is that not the LLM guy?

Posted May 11, 2026 14:49 UTC (Mon) by rahulsundaram (subscriber, #21946) [Link]

> * All code must be compatible with GPL-2.0-only

>> … which clearly excludes LLM-regurgitated slop.

Does it? I understand you wish it was so but so far there hasn't been any clear rulings that have indicated that. If anything it looks like the opposite situation in practice. Refer to https://lwn.net/Articles/1061534/