On Fri, 7 Aug 2026 at 11:03, Gerardo Ballabio <[email protected]>
wrote:
> Simon Richter wrote:
> > On 8/5/26 5:59 PM, Gerardo Ballabio wrote:
> >
> > > As I understand it, the problem with copyright isn't that AI providers
> might claim copyright. It's that *someone else* might claim copyright
> because the AI scraped and regurgitated their code. That's still an open
> legal question AFAIK.
> >
> > That, and whether the person prompting AI can claim copyright on the
> generated code.
>
> So if I ask a *question* I may claim copyright on the *answer*?
> Well, in a sense, it could be argued that the answer is a "derivative
> work" of the question. But I don't think that anybody would take this
> argument seriously.
> Certainly it doesn't work like that when you ask a human.
> Why would asking a machine be different?
>
It depends. If I ask a question - "how will this (attached) C file look
like with race conditions fixed?" then the answer will be a derived work of
the attached file, regardless of if the machine or a human answer that. If
I ask "please write a polite email in Spanish that would ask hotel to
prepare the sauna for our arrival at 22:00 on Thursday" then the answer is
just as derrived work of my question as a simple translation.
> Additionally, I'm pretty sure that the terms of service of every
> corporate LLM require users to surrender any rights they might have.
They don't. Otherwise they would loose all of their corporate customers,
where they are actually making most of their money. The legal departments
of corporate customers read every work of those agreements, highlight
problems to service providers and get problematic language fixed before
they allow the company to use a particular model or service. Grok xAI is
the only exception, which is why its use is banned in most corporate
contexts.
--
Best regards,
Aigars Mahinovs
