Re: LLM Commit in upstream `master`?!

because that's how Fable works, you ask for it to write a commit 
and it
is "trained" to sign the commit, then Yuan just let it pass by 
mistake
and that's how it ended in there.

On sáb, jul 25 2026, Divya Ranjan Pattanaik wrote:

> Hello Sean,
>
> But this begs the question: why would someone with commit access 
> try to edit/prepare a commit for Emacs when
> there is an explicit policy from refraining to use LLMs for 
> commits/patches?
>
> I can understand that Yuan might've forgotten and pushed the 
> commit by mistake, or their agent did so, but the
> fact that the agent had such permission in the first place, is 
> very troubling. If people with commit access are
> allowing agents to edit/prepare commits then it is a risk. I 
> don't think this should be desirable in the light
> of the current interim policy.
>
> Regards,
>
> On 25 July 2026 17:00:03 GMT+05:30, Sean Whitton 
> <[email protected]> wrote:
>
>  Christine Lemmer-Webber [25/Jul  7:09am -04] wrote:
>
>   I, too, am disturbed by this. I'm also not a fan of the way 
>   that
>  "Co-authored by Claude Fable 5" reminds me of "sent from my 
>  iPhone", it
>  ends up being a marketing system for these organizations.
>
>  On the fediverse it was also pointed out that the AuthorDate is 
>  listed
>  as May 20th 2026:
>
> https://goto.boserup.eu/@ah/statuses/01KYC7QYHVEEF2WD8ZHTFV7YDD
>
>  This seems impossible, "Claude Fable 5" was not even available 
>  for
>  public use by then.
>
>  What is going on here?
>
> Perhaps Yuan started working on it by hand in May and then the 
> LLM was
> used to amend the commit the later.
>
> I'm sure that this was just a mistake or misunderstanding.
>
> Yuan, I think we have to revert this commit, but before doing 
> so, can
> you explain what happened here?
>
> Divya Ranjan Pattanaik
>
> Mathematics, Philosophy & Libre Software

-- 
Jorge Araya

Contacto:
 Telegram: t.me/shackra · Signal: Shackra.28