[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Proposal -- Interpretation of DFSG on Artificial Intelligence (AI) Models



Hi Aigars,

At 2025-05-04T14:27:46+0200, Aigars Mahinovs wrote:
> The simple fact that none of the LLMs have been sued out of existence
> by *any *copyright owner is de facto proof that it does not work that
> way in the eyes of the judicial system.

I think that's a hasty conclusion.  Civil procedure is slow; lawsuits
can take several years to get started, and, in common law jurisdictions
like the U.S., the principle of laches (an analogue to the statue of
limitations for criminal procedure) can be foreclosed by all sorts of
non-public activities, like corporate attorneys jousting with each other
via private communications.  This can go on for a long, long time, and
one's standing as a civil plaintiff can be preserved by such efforts.

Further, one doesn't sue the LLMs themselves--a model can't be a party
to a lawsuit--or even necessarily the party that constructs one, but the
party that "monetizes" the result.  That latter for two good reasons:
(1) the monetizer would likely be the party that engaged in copyright
infringement--_if_ that concept applies--because they're the ones
engaging in public commerce and thus "distribution" of copyrighted
materials, and (2) the monetizer is likely to be party among all of
those involved with LLM construction that has the deepest pockets, and
money damages are the substrate language of most civil procedure.

You _might_ be right, and the more time that passes without a big
lawsuit hitting the newspapers, the more likely you are.  But I think
the hour is still young in this matter.

Not really relatedly, I wonder how the SFC v. Vizio case[1] is going...

Regards,
Branden

[1] https://sfconservancy.org/copyleft-compliance/vizio.html

Attachment: signature.asc
Description: PGP signature


Reply to: