On Fri, 2026-02-06 at 11:04 -0700, Nathan via NumPy-Discussion wrote:
> On Fri, Feb 6, 2026 at 10:57 AM Charles R Harris via NumPy-Discussion
> <
> [email protected]> wrote:
> > 

<snip>

> > 
> > I don't think we have a formal position at this point.
> > 
> 
> It’s probably time to adopt one and/or add an AGENTS.md file to the
> repo.
> 
> I mostly agree with Chuck that there’s mot much we can do to avoid
> it.
> People will use the tools and not disclose, so any copyright issues
> will
> happen no matter what policy we have.
> 
> I’m nervous about subtle inconsistencies and hallucinations,
> especially
> from contributions that are mostly vibe-coded. To me, that means the
> code
> needs much more careful review than human-written contributions
> because the
> nature of the errors made are different.
> 

I am just going to third this sentiment. There is a lot of interesting
discussions around this but at the core of it, to me there isn't
actually a real core policy shift yet. (Which doesn't mean it wouldn't
be good to adopt one.)

What I mean is that I think contributors always should explain where
code comes from (you can map this to copyright, or AI, or a colleague
attribution).
Also, I expect a contributor to understand the code they are proposing
as well as why they propose the change and, to some extend [1], what
impact that change has on backwards compatibility.

The same is true for the maintainer merging that change, giving us the
four eyes principle (you may stress that it is at least four human
eyes).
And that leaves me fine with contributions even largely written by a
tool so long the understanding is still there (and as a reviewer I may
need to know about the tool use, I think).
And I expect the project could lose a lot of user trust if the number
of human eyes on a changeset suddenly dropped. [2]

All that said. I am happy to be rather ruthless with suspect tool use
and closing of PRs/issues (especially if they don't disclose use).
If contributors lack understanding or don't take care they shift work
on maintainers and, unfortunately, a perceived increase in such
issues/PRs means I am fine if maintainers react rather harshly. [3]

- Sebastian


[1] This part is hard for many contributors, it indeed falls more on
the maintainer.
[2] I am not sure it matters how well it might work now or in the
future, because NumPy seems like the wrong project to trail-blaze
shifts in culture.
[3] I don't like the cultural implications of being trigger happy with
closing PRs, especially since it is hard to be sure about anything
here. But, unfortunately, more tool use may mean that contributors have
to do actually do more work to make PRs nice to review.


> 
> > Chuck
> > _______________________________________________
> > NumPy-Discussion mailing list -- [email protected]
> > To unsubscribe send an email to [email protected]
> > https://mail.python.org/mailman3//lists/numpy-discussion.python.org
> > Member address: [email protected]
> > 
> _______________________________________________
> NumPy-Discussion mailing list -- [email protected]
> To unsubscribe send an email to [email protected]
> https://mail.python.org/mailman3//lists/numpy-discussion.python.org
> Member address: [email protected]
_______________________________________________
NumPy-Discussion mailing list -- [email protected]
To unsubscribe send an email to [email protected]
https://mail.python.org/mailman3//lists/numpy-discussion.python.org
Member address: [email protected]

Reply via email to