Linus Torvalds to critics of AI coding in Linux: "Fork it. Or just walk away."
Ars Technica Kyle Orland ● Covered by 2 sources
Linus Torvalds told Linux kernel critics to fork the project or walk away if they hate AI coding tools. It's a blunt rebuke to open-source advocates pushing for an outright LLM ban.
Based on reporting by Ars Technica, Kyle Orland — read the original for the full story.
Summary, retelling and take written by AI under human oversight; images are AI-generated illustrations. How we work · Report an error
Linus Torvalds doesn't do subtlety, and this week's post to the Linux kernel mailing list proves it once again. Responding to a heated thread about AI-assisted code review, the kernel's creator declared that Linux "is not one of those anti-AI projects" and that anyone who disagrees can fork the codebase or simply leave. No hedging, no committee vote — just Torvalds putting his foot down, as he put it himself.
The fight erupted over Sashiko, described as an agentic Linux kernel code review system. Its creators claim that in testing it independently caught 53.6 percent of the bugs that human coders eventually fixed in later commits. That's a real number worth sitting with, because it suggests the tool is catching more than half of what maintainers would otherwise find themselves, sometimes much later. But Sashiko isn't flawless. Its own maintainers estimate the false-positive rate sits somewhere within a 20 percent range, meaning a chunk of its bug reports point to problems that don't actually exist.
That tradeoff is exactly what set off the mailing list argument: should maintainers be forced to sift through a wave of automated bug reports, some real and some noise, generated by an AI system nobody asked for? One poster invoked the Software Freedom Conservancy's recent statement, which argues that open source communities should "support, not just tolerate" contributors who outright reject LLM-generated AI systems, and that every FOSS contributor deserves self-determination on the matter.
Torvalds wasn't having it. He rejected the idea that any contributor gets veto power over whether the entire project accepts LLM-generated code or revisions. His stance, in his own words, is that nobody is being forced to use these tools — but he will "very loudly ignore" anyone trying to argue that other people shouldn't be allowed to use them either. It's a distinction that matters: this isn't Torvalds mandating AI tooling for every contributor, it's him refusing to let a vocal minority set a project-wide ban.
The episode captures a fault line running through open source right now, and Linux, given its size and influence, was always going to be a proving ground for it. Sashiko's imperfect but not-nothing bug-catching record is a decent stand-in for the whole debate: good enough to be useful, flawed enough to be controversial, and apparently good enough for the guy who runs the kernel to defend it in public.
My take — AI-written commentary, not fact-checked reporting
Torvalds picked the only position that actually scales for a project the size of Linux: nobody gets a personal veto over what tools thousands of other contributors are allowed to use. A 20 percent false-positive rate on an AI review tool is genuinely annoying for maintainers sorting through their inbox, and that complaint is legitimate. But turning that annoyance into a demand that the whole project reject LLM tooling outright is a different ask entirely, and it's the one Torvalds correctly refused to entertain. Open source has always run on forking as the pressure-release valve for disagreements like this — so the people who hate AI-assisted review already have their answer, however unsatisfying it sounds.'
Read more about this at: Ars Technica
