Linux kernel now accepts AI‑generated patches — but you wear the consequences

What's new

It has been reported that the Linux kernel project's documentation on GitHub now allows contributors to use AI to generate code so long as the output complies with existing submission guidelines, licensing, and attribution requirements. The guidance — first flagged by community chatter and covered in XDA Developers — permits AI-assisted contributions but draws a hard line around legal certification: "AI agents MUST NOT add Signed-off-by tags." Only a human can sign the Developer Certificate of Origin (DCO).

The rules — and the risk

In plain terms: you can copy in that snippet your model spat out, but you are the author in the eyes of the project. Human submitters must review every line of AI-produced code, ensure it meets license obligations, add their own Signed-off-by tag, and accept full responsibility for the contribution. If the patch breaks the kernel, introduces licensing problems, or otherwise causes trouble, the blame stops with the person who submitted it — not the assistant. No pointing the finger at Claude Code this time.

Why it matters

This is a pragmatic middle ground amid a patchwork of responses to AI across open source: some projects ban it, others lean in. The kernel's stance is cautious and enforceable — trust the tool, verify the output, and sign for it. For contributors, that’s both freeing and sobering: AI can speed development, but it doesn’t erase legal or technical accountability. Will this reduce friction or just shift more review burden onto maintainers? Time — and a few broken builds — will tell.

Sources: xda-developers.com