Does Debian now require contributors to say when they used AI?

Disclosure is encouraged but not required under Debian's new policy. The winning option asks contributors to be open about AI assistance without making a declaration mandatory, which is what separates it from the competing proposal that would have forced disclosure on every AI-assisted contribution.

Can a Debian reviewer reject a patch just because AI helped write it?

The resolution says Debian neither endorses nor prohibits generative AI tools, and sets no special rule beyond the standards every contribution already has to meet. A patch is judged on quality, correctness, maintainability and legal compliance, so "an AI wrote part of it" is not on its own a reason to refuse the work.

What happened to the proposals that wanted to ban AI in Debian?

All of them lost. LWN reports the two options that would have rejected AI-assisted work ranked below "None of the Above". The option titled "Debian is created by humans" lost to Responsible Use of Generative AI by 251 to 139, and the option to accept AI work only for Debian-specific tasks lost 232 to 115.

Can a Debian contributor paste private project data into ChatGPT or Claude?

Not without permission. Alongside the accountability rules, the Debian policy says contributors must not share confidential project data with third-party AI services unless they have explicit permission to do so, so the usual handling rules for private data survive the change.

Debian Project · 2026-08-28 · major

Debian allows generative AI — contributors stay responsible for the code

Debian developers picked "Responsible Use of Generative AI" in a general resolution that closed on August 28. It beat the closest rival 203 to 148. AI tools are neither endorsed nor banned, and the contributor still owns the result.

Phoronix

Debian's developers voted to allow generative AI in contributions, and to keep the person who submits the work fully on the hook for it.

Quick facts

Project	Debian
Decision	Responsible Use of Generative AI
Voting period	August 15–28, 2026
Method	Condorcet, Schwartz set
Closest margin	203 to 148
Disclosure	Encouraged, not required
Debian private data	Needs permission before it goes to an AI service

What is it?

Debian now has a written position on generative AI: contributions made with AI tools are allowed, and the contributor carries the same responsibility for them as for anything else they submit. The general resolution ran from August 15 to August 28, 2026, and the winning option was "Responsible Use of Generative AI". It asks contributors to understand, review, test and, where needed, change AI-assisted output before it goes into the distribution.

How does it work?

Eight options were on the ballot, from a full ban on LLM contributions to a proposal built on climate concerns, and Debian counted them with its Condorcet method. Responsible Use of Generative AI was the only option in the Schwartz set, so it beat every other choice head to head — 203 to 148 over the option that would have allowed AI work with extra conditions, 210 to 130 over the cautious option, and 251 to 139 over "Debian is created by humans".

Why does it matter?

Debian sits under Ubuntu and thousands of downstream systems, so a distribution of that size settling the question gives its maintainers a rule to point at instead of a per-thread argument. A maintainer who avoided AI tooling because the policy was unclear can now use it, and a reviewer can no longer refuse a patch purely on the grounds that a model helped write it. The limits are still real: contributors own the licensing and correctness of what they submit, and Debian's private data stays out of third-party AI services without permission.

Who is it for?

Debian developers, package maintainers and downstream distributions

Frequently asked questions

Does Debian now require contributors to say when they used AI?: Disclosure is encouraged but not required under Debian's new policy. The winning option asks contributors to be open about AI assistance without making a declaration mandatory, which is what separates it from the competing proposal that would have forced disclosure on every AI-assisted contribution.
Can a Debian reviewer reject a patch just because AI helped write it?: The resolution says Debian neither endorses nor prohibits generative AI tools, and sets no special rule beyond the standards every contribution already has to meet. A patch is judged on quality, correctness, maintainability and legal compliance, so "an AI wrote part of it" is not on its own a reason to refuse the work.
What happened to the proposals that wanted to ban AI in Debian?: All of them lost. LWN reports the two options that would have rejected AI-assisted work ranked below "None of the Above". The option titled "Debian is created by humans" lost to Responsible Use of Generative AI by 251 to 139, and the option to accept AI work only for Debian-specific tasks lost 232 to 115.
Can a Debian contributor paste private project data into ChatGPT or Claude?: Not without permission. Alongside the accountability rules, the Debian policy says contributors must not share confidential project data with third-party AI services unless they have explicit permission to do so, so the usual handling rules for private data survive the change.