GCC will decline legally significant contributions made with AI models, except test cases

Sources: gcc mailing list, July 2026, Phoronix GNU index
Summary: The gcc mailing list archive shows a GCC AI Policy Announcement thread opened by David Edelsohn on 2026-07-29, with replies. The announcement body could not be read from this run environment, because pipermail message URLs are not recoverable through a link-stripped fetch. The policy detail carried here, that the Steering Committee accepted the GCC AI Policy Working Group recommendation to decline legally significant contributions produced with AI models while allowing them for test cases, rests on Phoronix reporting alone and is not verified against the announcement text. The Phoronix GNU section index carries that item dated 2026-07-29 under the headline that GCC will decline significant contributions made via AI or LLMs except for test cases. The GCC contribution page was checked on this run and still carries no AI clause, last modified 2026-07-23.
Why it matters: A copyright-provenance rule at the compiler propagates into every downstream contribution workflow, and GCC is the second large toolchain project this month to move on the question, after Debian opened a general resolution on LLM usage.
Follow-up: Re-verify the policy against the announcement message once a readable URL is available.