GCC To Decline Any Significant Contributions Made Via AI/LLMs - Except For Test Cases

The GCC Steering Committee announced that they have accepted the recommended policy of the GCC AI Policy Working Group, which means no “legally significant” code contributions made via AI/LLM agents will be accepted into this open-source compiler.

Key takeaways from the policy:

“For the time being, the GNU Compiler Collection (GCC) policy is to decline any legally significant contributions which include LLM-generated content or are derived from LLM-generated content. … The GCC maintainers are free to accept legally insignificant contributions generated by an LLM as long as they meet the usual prerequisites for any contribution and the contribution is clearly marked. … As an exception, the GCC maintainers are free to accept legally significant contributions to test cases, generated in whole or in part by an LLM.”

This largely aligns with the existing GNU policies of blocking AI/LLM contributions. The GCC stakeholders plan to revisit the policy in early 2027.

Additional context (LWN / Linuxiac / It’s FOSS)

• “Legally significant” follows GNU maintainer guidelines: around ~15 lines of code and/or text for copyright purposes
• Policy does not forbid using LLMs for research, analysis, bug discovery, reporting, or patch review — as long as LLM output is not included in the contribution
• Derived/edited LLM code still falls under the ban; contributors cannot launder AI-generated patches as human-written
• AI-assisted contributions that are accepted as insignificant should be marked (e.g., Assisted-by: tags, similar to Linux kernel practice)
• Announced ~July 29, 2026; sits between Codeberg’s broader AI-repo ban and LLVM’s disclosure-oriented approach