Working Group for GCC AI Policy
This document is a work in progress
This document presents information that is not yet finalized and still under discussion
This is the WG for the development of an initial AI policy for the GNU Compiler Collection (gcc).
Membership
The GCC SC approved the WG on 2026-04-24 with Jonathan Wakely as the lead.
The Lead for the working group is responsible for gathering volunteers to carry out the working group goals.
- Lead: Jonathan Wakely (lead)
- Volunteers:
- Carlos O'Donell (volunteering in cross-project capacity with interest in a harmonized GNU Toolchain AI policy)
- Sudakshina Das
- Jason Merrill
Problem Definition
Contributors want to know if they can use AI to develop all or part of the work they contribute to the project.
The GNU Compiler Collection (gcc) should have a policy on the use of AI technology to give contributors appropriate guidance.
The term AI generally encompasses all technologies currently known as "Artificial Intelligence" including small language models (SLMs) and large language models (LLMs).
Background on the Problem
The GNU Toolchain has started to receive contributions from developers using LLMs to create all or part of the contributions.
Other projects in the ecosystem like the Linux kernel, QEMU, and even whole distributions have already published preliminary AI policy.
Goals
The goal of the working group is to draft an initial preliminary recommended AI policy for gcc for the GCC SC to review to adopt for the project.
The assessment and justification behind the policy are as important as the policy itself, and as such a secondary goal is a detailed assessment section with external references.
Lastly, a goal of the working group should be to decide if a simpler initial policy should be released immediately while a more detailed policy is drafted by another working group. For example a simpler initial policy can clarify that assistive technologies based on LLMs are permitted.
The timeline for the initial assessment should be no more than 3 months.
Ways of Working
TODO
Assessment
TODO: WG to write up the assessment of the problem.
Recommendation
TODO: WG to write up a recommendation.
Status
- 2026-04-16: Created Draft working group on AI policy
2026-04-24: Approved by the GCC SC.
- 2026-04-27: Update WG page with next steps.