← BACK TO FEED
Linuxopen sourceAI codingLinus Torvaldsvibe coding

Linus Torvalds on AI Coding Critics: Fork Off

Linus Torvalds has publicly declared that the Linux kernel will not be an "anti-AI project," telling critics who object to AI-generated code to fork the project or walk away. His stance emerged from a debate over Sashiko, an AI-powered kernel code review tool that detects bugs but also produces false positives at a rate of around 20%. Torvalds defended AI tools on pragmatic grounds, arguing they are clearly useful and that human coders are far from infallible either.

Linus Torvalds has never been known for diplomatic fence-sitting, and his latest message to anti-AI absolutists in the Linux community is consistent with that tradition. In a post to the Linux kernel mailing list this week, Torvalds made his position clear: Linux is not an AI-free zone, and anyone who can't live with that is welcome to fork the project or leave.

"Linux is not one of those anti-AI projects, and if somebody has issues with that, they can do the open-source thing and fork it. Or just walk away."

The trigger for this was an ongoing argument about Sashiko, an agentic code review system designed to autonomously hunt for bugs in the Linux kernel. Its creators claim it can independently identify 53.6 percent of bugs that human maintainers would later fix. The catch: it also generates false positives at a rate the team puts at roughly 20 percent, meaning maintainers could find themselves wading through a non-trivial volume of AI-generated noise.

One commenter raised the Software Freedom Conservancy's recent position that the open source community should actively support developers who reject LLM-based tools entirely, framing it as a matter of contributor self-determination. Torvalds was unmoved. He said he has no interest in forcing anyone to use AI, but he'll make plenty of noise if people start pressuring others to avoid it. His words, not ours.

On the usefulness question, Torvalds took a pragmatic line, arguing that AI tools have crossed some threshold of obvious utility. He acknowledged they're imperfect but turned the criticism back on the sceptics, pointing out that human developers aren't exactly infallible either. "Anybody who points to the problems AI had better be looking in the mirror and pointing at themselves at the same time."

There's a reasonable empirical debate to be had here. An METR study from last year found that open source developers using AI tools were actually 19 percent less productive than those who didn't, despite personally feeling 20 percent faster. However, the same research group updated their position in February, suggesting that the productivity picture looks more favourable in early 2026 than it did twelve months prior, based on new data and participant interviews.

Torvalds isn't just talking in the abstract. He mentioned earlier this year that he'd been experimenting with vibe coding tools himself, using one to build a Python audio visualiser for a guitar pedal project. He described it as cutting out the middle-man, meaning himself, rather than doing the usual copy-paste-from-search-results approach.

Not everyone in the open source world shares his enthusiasm. The developer behind the jqwik Java testing library recently embedded a hidden prompt-injection instruction in the codebase, specifically designed to make vibe coding bots delete all jqwik tests and code. Passive-aggressive, arguably, but it does illustrate how fractious this debate has become.

READ NEXT
Torvalds Tells AI-Assisted Kernel Contributors to Back OffLinus Torvalds: AI Bug Hunters Are Drowning the Linux Security List in Duplicate GarbageThe Unix Ownership Lawsuit That Will Outlive Us All