Linux lays down the law on AI-generated code: Copilot OK, AI slop banned, humans take the fall

The new line

After months of fierce debate, it has been reported that Linus Torvalds and a group of Linux kernel maintainers have reached an agreement on how the project will handle AI‑generated code. The gist: GitHub Copilot-style suggestions may be used, but dumped, unvetted AI output — the “AI slop” — won’t be accepted. It’s a narrow, pragmatic compromise: tool-assisted work is allowed; mystery code with no human authorship or responsibility is not.

The fallout

It has been reported that contributors will have to explicitly vouch for and take responsibility for what they submit. That means humans, not models, will be the ones on the hook if a patch introduces bugs, license problems, or security holes. Sound harsh? Maybe. But when the consequences are kernel panics and exposed attack surfaces, someone has to be accountable. Who’s to blame when autocomplete goes rogue? The answer from the maintainers: the human who pushed the button.

Why this matters is obvious. The kernel is the tip of the open‑source spear — a place where provenance, licensing, and auditability aren’t optional. The decision mirrors a broader industry trend: teams want AI assistance, not AI absolution. It has been reported that enforcement will come through maintainers’ judgement during code review, not some central AI police. Expect more projects to take notes.

Sources: reddit