glibc AI policy

[Carlos O'Donell]

On 4/5/26 10:35 AM, Jeffrey Law wrote:
> A de-minimis contribution is (at least in the US) considered so
> trivial that is not copyrightable, so I'd tend to suspect such
> contributions, even from LLMs, wouldn't be a problem under the FSF's
> de-minimis guidelines of 5 lines of code.  Of course de-minimis
> contributions in aggregate may cross the threshold into
> copyrightable.
Just for clarity the current guidance is documented here:
https://www.gnu.org/prep/maintain/maintain.html#Legally-Significant

The guideline is ~15 lines of code.

The rule exists because very small changes are often considered
"de-minimis" in copyright law and not creative enough e.g. typo,
variable rename, etc.

So long as the changes are legally insignificant we could apply them
to the GNU Toolchain sources without consideration of copyright?

Also note the guideline says that it could be *more* than 15
lines if the change is mechanical in nature e.g. repeated, renaming,
many places etc.

There isn't a hard and fast rule here, but when a contribution
crosses the boundary into legally significant the question of
copyright assignment or DCO should be raised,

Lastly, Richard Fontana is publicly on the record here saying that
DCO can be used to attest to LLM contributions:
https://www.redhat.com/en/blog/ai-assisted-development-and-open-source-navigating-legal-issues
~~~
With disclosure and human attentiveness – and oversight – aided
where possible by tools that check for code similarity, AI-assisted
contributions can be entirely compatible with the spirit of the DCO.
~~~

Unfortunately this has several caveats to consider.

I have been using Claude Code extensively for patch review,
and if we had a minimal policy that clarified this usage, like the
usage for assistive technologies, I think it would help contributors
to the project know what we would and would not accept *today*
(even if that changes *tomorrow*).

-- 
Cheers,
Carlos.