Re: GCC AI Policy Announcement

> Am 29.07.2026 um 17:30 schrieb Arthur Cohen <[email protected]>:
>=20
> =EF=BB=BFHi Richard,
>=20
> What are the important LLM issues that you think the policy fails to addre=
ss? As the policy itself states, the it is expected to evolve and feedback i=
s welcome. The goal is to continuously improve it.

The policy grounds its rejection of legally significant LLM contributions on=
 legal issues (by means of naming).  It does not cover the IMO much more sev=
ere issues of overwhelming reviewers and keeping GCC maintainable.  It also m=
akes LLM assisted contributions by maintainers who might want to use LLMs to=
 get to prototype stages fall into grey areas or even rejected.  I realize t=
he policy is not a discussion review document but what was the point of havi=
ng the policy group when the outcome is just the obvious =E2=80=9Alegally un=
clear situation=E2=80=98 statement?

Richard=20

> Thanks,
>=20
> Arthur
>=20
>> On 7/29/26 9:06 AM, Richard Biener wrote:
>>> On Wed, Jul 29, 2026 at 8:07=E2=80=AFAM Teylu, Andrew via Gcc <[email protected]=
nu.org> wrote:
>>>=20
>>> Long time listener, first time caller =E2=80=A6
>>>=20
>>> For those of who don=E2=80=99t click through, the rough summary is: to d=
ecline any patches longer than 15 lines that are written by AI.
>>>=20
>>> Is anyone else surprised by this? When the threads first started on this=
 topic earlier on, I had believed that GCC was =E2=80=9Cmore optimistic=E2=80=
=9D on the use of generative AI, but this new policy seems rather =E2=80=9Ca=
nti AI=E2=80=9D (a change of only 15 lines seems very limiting).
>>>=20
>>> To be clear, I am not trying to get the policy changed, I am just wonder=
ing if any other developers are surprised at this?
>> I'm not surprised that we take a conservative stance on copyright
>> issues related to LLM contribution.  I'm somewhat
>> surprised we're leaving open the grey area of "legally insignificant"
>> contributions.  The cited 15 lines is not a hard
>> number and is not directly referred to in the policy.
>> In the end the policy side-steps all important issues with LLM
>> contributions by deflecting to legal issues ...
>> Richard.
>>> From: Gcc <[email protected]> on behalf o=
f David Edelsohn via Gcc <[email protected]>
>>> Date: Wednesday, 29 July 2026 at 03:12
>>> To: GCC Development <[email protected]>; Jonathan Wakely <jwakely.gcc@gmai=
l.com>
>>> Subject: GCC AI Policy Announcement
>>>=20
>>> The GCC Steering Committee has accepted the recommended policy of the GC=
C
>>> AI Policy Working Group.  The policy is described in the following docum=
ent
>>> preview / pull request in Forge, which will be committed to the GCC webs=
ite:
>>>=20
>>> https://forge.sourceware.org/redi/gcc-wwwdocs/commit/4d0793a6a14bf9bfe9e=
92ac1599840780355199d
>>>=20
>>> We expect the policy to continue to evolve and will revisit it periodica=
lly.
>>>=20
>>> We wish to thank Jonathan Wakely for his leadership of the working group=

>>> and the members of the working group: Carlos O'Donell, Sudakshina Das, J=
ason
>>> Merrill, Joel Sherrill, Sam James, Robin Dapp, and Arthur Cohen.
>>>=20
>>> Happy Hacking!
>>> David
>=20
> <OpenPGP_0x1B3465B044AD9C65.asc>