Matthias Geiger [22/Jul  5:35pm +02] wrote:
> 1. Copyright
>     LLM output has very unclear legal status: it may be possible to copyright 
> on its own merits, or not; it may be affected by all of the licenses and 
> copyrights in the training data, or not.
>     Debian Policy and the DFSG require absolute clarity for licensing and 
> copyright[1][2]. Software and other contributions written conventionally by 
> humans with unclear copyright or license
>       status are not allowed in Debian;
>     LLM output should not have a special exception to this.

The FSF are currently consulting with lawyers to write an LLM
contributions policy for GNU.  These concerns are their overriding
focus.  Things may become much clearer once we see the result of that
work, and so I think that we should wait to see it before making our own
statement on the matter.

>     Other categories
>     such as upstream projects written with LLM assistance may be included at 
> a later date.
>     This ensures that Debian remains a stable, trusted, and reliable operating
>     system, and protects the interests of the Debian volunteers who make it 
> possible.

So many[1] core things already have LLM contributions, so I think "at a
later date" is would make a big chunk of the archive RC-buggy
immediately.  I don't think we should impose that burden on package
maintainers.  So I suggest this this is not left to a later date.

Finally, as I've said before, I think we should scope this to
"LLM-generated contributions" so that we are not disallowing people
using LLMs to, e.g., root cause a problem without generating a fix.

I understand that you and others would prefer people don't do that
either, but it makes the proposal one that more people can agree to.

[1]  https://git-annex.branchable.com/no_llm_code/

-- 
Sean Whitton

Attachment: signature.asc
Description: PGP signature

Reply via email to