Re: GCC AI Policy Announcement

--=-=-=
Content-Type: text/plain

Thomas Koenig via Gcc <[email protected]> writes:

> I think the new AI policy is too restrictive, and does not
> need to be so to serve the purpose of the GPL.
>
> If you read the preamble of the GPL v3, you will find
>
> # the GNU General Public License is intended to guarantee your freedom
> # to share and change all versions of a program--to make sure it remains
> # free software for all its users.
>
> and
>
> # Developers that use the GNU GPL protect your rights with two steps:
> # (1) assert copyright on the software, and (2) offer you this License
> # giving you legal permission to copy, distribute and/or modify it.
>
> so copyrighting the software is a means to an end, not an end in itself.
>
> What happens if a patch contains a non-trival amount of code that
> cannot be copyrighted because it was assisted by AI?

The open questions around this are why the GNU project issued
provisional guidance to reject significant contributions developed with
an LLM.

If it is copyrightable, then who owns the copyright? If it's based on
the training data, then the contributor may not have the rights to
share such a patch, and it cannot be legally redistributed under the GPL
(or any licence).

>
> If some third party were to distribute a binary of gcc, they would
> still be obliged to distribute the source that is covered by the GPLv3.
> Reverting the non- copyrightable patches before source distribution
> would break things and make no sense. Alternatively, distributing
> only the copyright-free patches would also make no sense, because
> they would not result in a working compiler.

The point is to not allow such patches into the repository to begin
with, so that such a problem doesn't happen?

>
> A gcc version which contains both copyrighted and non-copyrighted code
> would therefore still be protected.
>

This assumes it isn't copyrightable, though.

But we didn't assume that it is, or isn't, copyrightable. We chose a
conservative position because it's not clear.

> I would therefore suggest to focus on the practical aspects that Richard
> mentioned, such as the danger of architecturally poor-quality code,
> potential reviewer overload etc.

It's not clear to me that Richard was advocating neglecting the
copyright side, rather wanting other issues to be addressed as well.

>
> Best regards
>
> 	Thomas

sam

--=-=-=
Content-Type: application/pgp-signature; name="signature.asc"

-----BEGIN PGP SIGNATURE-----

iQEBBAEWCgCpFiEEJaa7iN2bdkxrVUHCc4QJ9SDfkZAFAmpuQlwbFIAAAAAABAAO
bWFudTIsMi41KzEuMTIsMiwyXxSAAAAAAC4AKGlzc3Vlci1mcHJAbm90YXRpb25z
Lm9wZW5wZ3AuZmlmdGhob3JzZW1hbi5uZXQyNUE2QkI4OEREOUI3NjRDNkI1NTQx
QzI3Mzg0MDlGNTIwREY5MTkwDxxzYW1AZ2VudG9vLm9yZwAKCRBzhAn1IN+RkD3V
AQCbsHOPmpWyLfPZGYYyGQtexZr+wY4pgVHL5Cxqh1SCIQD9ErIcutuqQIBuxEMz
d8obSLFtgPavD75/RC+W4mx2ewQ=
=1NQu
-----END PGP SIGNATURE-----
--=-=-=--