Hi folks,

Handling AI contributions is an area we need to talk more about at some
point. Some other ASF projects receive 10 or more AI-generated "security
fix" proposals on some days. We certainly aren't impacted by this just yet,
but we should prepare in case we see a rise in such PRs.

Also, the ASF has some general advice on generative coding:

https://www.apache.org/legal/generative-tooling.html

A key part of that advice is that humans need to vouch for what gets
submitted/approved. However, we can take steps to ensure future submissions
meet minimal requirements if AI is involved somewhere in the process.

So, I have put together an initial version of an AGENTS.md file that
attempts to capture at least the "comply with ASF licensing part" of the
general advice. We should undoubtedly expand the AGENTS.md file, but there
is a lot of activity happening within and outside the ASF in this space, so
we don't need to rush.

Please discuss if you have any thoughts/concerns.

Cheers, Paul.

Reply via email to