Hello Jean.

Fully agree on this.

My personal opinion is that we should expect the same level of quality and 
engagement in the review process that for humans PRs. We should expect the 
human to be accountable for the code written by the LLM, able to explain it &al.
If the criteria are not met, maintainers shall be able to disregard the PR 
after an observation period of 3 days, pointing to the above notice.
We should also expect contributors to state the exact use of LLM when used (eg 
used to generate code skeleton that I heavily reviewed and improved).

Regarding the ASF there is ongoing (firy) discussions with mixed opinions, 
including intellectual property concerns. IMO not an issue as long as the human 
author did put enough energy to make this an original work. And not ok if the 
thing had been vibe coded away.
The overall consensus emerging in "My Current Experience and Feeling on Agentic 
Coding" member discussion seems to be "ASF is to make sure that the AI makes 
"us" more productive and deliver better products" which is quite in line with 
this.-- 

Best regards,

Benoit TELLIER

General manager of Linagora VIETNAM.
Product owner for Twake-Mail product.
Chairman of the Apache James project.

Mail: [email protected]
Tel: (0033) 6 77 26 04 58 (WhatsApp, Signal)


Le févr. 27, 2026 8:04 AM, de Jean Helou <[email protected]>Hello,

Since I keep seeing mentions of conflicts arising in projects without a
clear policy when contributions made with LLM assistance are proposed to
projects, I think we should establish a clear policy in advance.

I'm also not quite sure what the Apache foundation stance on this is, I
haven't found or heard of an official foundation wide position but I may
have missed it. The intellectual property risk associated with LLM usage in
open source projects is unclear to me.

What do you think ?

jean

Reply via email to