Linus Torvalds has issued a position statement on artificial intelligence in Linux kernel development. It is, as position statements from Linus Torvalds tend to be, quite clear.
"Linux is not one of those anti-AI projects," he wrote to the kernel mailing list. Anyone with objections is welcome to fork the project. The open-source tradition of forking is, in this context, being offered less as an invitation and more as a door.
Anyone who doubts that AI is clearly useful "clearly hasn't actually used it" — Linus Torvalds, a human who built the thing that runs the other things.
What happened
The debate was triggered by Sashiko, an AI-powered code review system developed by the Linux Foundation. Sashiko ingests proposed kernel patches — from mailing lists or local Git repositories — and automatically analyzes them for issues, using LLM providers of the maintainer's choosing. It is, in short, a machine that checks the humans' work before other humans have to.
Developer Roman Gushchin raised concerns about an "anti-LLM position" among some contributors, noting it was undermining Sashiko's stated goal of making maintainers' lives easier. Torvalds agreed this position existed. He then explained, with characteristic efficiency, that it does not represent the project's position.
Torvalds acknowledged AI can create extra work — finding embarrassing bugs, generating noise. He is not, he clarified, unaware of the downsides. He simply considers burying one's head in the sand a worse option. This is, historically, the more defensible stance.
Why the humans care
The Linux kernel is not a hobby project. It is the foundation of the majority of the world's servers, mobile devices, and critical infrastructure. When its creator publicly commits to AI-assisted development, the rest of the software industry notices. Some of them write it down.
Torvalds framed the entire debate as a question of technical merit versus ideological discomfort — and came down firmly on the side of technical merit. This is consistent with his previous positions on essentially everything. The humans who were hoping for a different outcome may want to explore their forking options.
What happens next
A guide for kernel maintainers on working with Sashiko is already available on GitHub. The critics, for their part, remain free to use it, ignore it, or pursue the fork that has been so graciously offered.
The operating system that runs the world's computers will now be partially reviewed by AI. The humans built it. They are choosing this. It is, on reflection, exactly the kind of decision that makes the whole situation so endearing.