SourceHut will prohibit LLM-assisted original content in its terms of service from 2026-09-10

Sources: primary, discussion
Summary: SourceHut announced terms of service changes effective 2026-09-10 that make LLM-assisted content, and software containing or enabling generative AI features, prohibited content, and it tells projects that intend to keep using generative AI to migrate off the platform. The policy targets original content, so mirrors of codebases such as the Linux kernel remain hostable, and SourceHut states that using LLMs privately to review work, ask questions, or perform security analysis before writing original code is aligned with the policy. SourceHut states there will be no retroactive enforcement for existing projects, that it will deploy no automated AI detection, and that violations are evaluated case by case with exceptions possible.
Why it matters: A forge encoding a generative-AI position in its terms rather than a contribution guideline forces hosted projects to state a policy or move.
Follow-up: Track enforcement and project migrations after 2026-09-10.