Debian votes to allow 'responsible' AI-assisted contributions
The winning framework keeps humans on the hook for every AI output — and a ~30% bloc still voted to restrict or ban it outright.
Copy markdown
Choice 5 wins: neither endorsed nor banned
Debian's General Resolution adopted "Responsible Use of Generative AI," which neither blesses nor bans LLM tools in code and docs, saying they can "substantially improve" contributor productivity when used responsibly. Voting closed Aug 28; results posted Aug 29.
The strings attached: you own the output
Contributors stay fully accountable — they must understand, review, test, and modify any AI-assisted output, and it must clear Debian's existing quality, correctness, and legal-compliance bar. Feeding secrets, credentials, or security-sensitive data to third-party AI services without authorization is off-limits.
Mass AI patches need a human green light
Large-scale automated actions — mass bug filings, cross-package patches, sweeping automated changes — now require prior community discussion and human oversight. Disclosing that you used AI is encouraged but not mandatory.
A third still wanted it restricted
The hardline options — an outright LLM ban, amending the Social Contract or Code of Conduct, or citing climate cost as a dealbreaker — all ranked below "None of the Above." But roughly 30% of developers put a stricter stance first, so the FOSS fight over AI code is far from settled.
Why it lands on your desk
Debian is a bellwether for open-source norms. A clear "AI-assisted patches are fine if you own them" ruling from one of the oldest distros hands other projects a ready template — and tells you the ground rules before you send AI-drafted contributions upstream.