If all goes well, Mesa developers are hoping to reach a consensus or at least some common ground on an AI policy in March. Mesa is the latest open-source project making considerations around the growing activity around AI coding agents and the like and how to deal with them for this project that is crucial to the Linux desktop and open-source 3D graphics drivers at large.

Last year [1]Mesa began making some movement around AI policies while this week Karol Herbst has been working on a more formalized AI policy for the project. The expressed hope is that some consensus could potentially be reached in the coming weeks otherwise continuing with the status quo.

Among the proposals being considered over the status quo are to disallow any use of autonomous AI agents, no substantial AI generated code, a complete AI ban, or full AI transparency. Alternatively, there could be different rules within the Mesa codebase depending upon the driver/component such as on a per-directory basis whether AI is okay or not.

For the most part the usual concerns raised by open-source projects at large and whether to allow AI or not and if so the level of transparency or whether significant code contributions can be made. The one unique proposal is potentially allowing AI on a per-project/directory basis within Mesa. That could quickly get messier to manage and acknowledged as being a sub-optimal result but a possibility if no global policy can be agreed upon.

Those curious can see the current draft of these different AI policy proposals for Mesa within [2]this merge request .



[1] https://www.phoronix.com/news/Mesa-Contributor-Guidelines

[2] https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/40109/diffs?commit_id=0c1a286f2aea094f6a4709b160632ee1403db596