# Linux内核确立AI辅助编程政策：开发者须对AI代码承担全责

- 来源：Rohan Paul (@rohanpaul_ai)
- 发布时间：2026-04-13 04:31
- AIHOT 链接：https://aihot.virxact.com/items/cmnw81tct008lslcjpllgsxwt
- 原文链接：https://x.com/rohanpaul_ai/status/2043426697027228026

## AI 摘要

Linux内核项目本周正式确立政策，允许开发者使用AI辅助编写代码，但须遵守严格的信息披露要求。Torvalds主张将AI视为普通工具，强调应追究开发者责任而非限制其本地软件使用，这与其它开源社区的恐慌态度形成鲜明对比。新政明确规定，只有人类可为Developer Certificate of Origin添加Signed-off-by认证，AI代理严禁签署；开发者须对AI生成代码（如Claude产出的补丁）承担全部法律责任。此举旨在应对当前开源社区"AI slop"泛滥的乱象。

## 正文

This week， the Linux kernel project finally created a formal， project-wide policy explicitly allowing AI-assisted code contributions， as long as developers obey strict new disclosure requirements.

Torvalds' view， which gives this policy its main philosophical shape， is pretty direct： AI is just another tool. Developers submitting garbage code are not going to be fixed by more documentation， so the kernel should hold people accountable instead of trying to control the software they use on their local machines.

It is a practical and reasonable line to take， especially compared with the panic in other parts of the open-source scene.

You are the one on the hook now. If Claude introduces for example， a race condition in the block layer and you approve it， the patch carries your tag， not the model's.

The Signed-off-by line is the certification for the Developer Certificate of Origin， and the latest policy makes it explicit that only humans can legally add it. AI agents "MUST NOT"

The open-source community is currently getting overwhelmed by what people are calling "AI slop."

e.g. the creator of cURL closed bug bounties after a flood of hallucinated code， tldraw began automatically closing external PRs to defend itself， and projects such as Node.js and OCaml have seen huge， >10，000-line AI-generated patches