The Linux kernel mailing list is a place where technical debates get resolved through sheer force of logic, and occasionally, the sheer force of Linus Torvalds’ personality. This week, the debate landed on AI tools in kernel development, and Torvalds didn’t just wade in. He nuked the conversation from orbit.
The message was unambiguous: Linux is not anti-AI, AI is a useful tool, and if you have a problem with other people using it, you can fork the project or walk away. No quarter given, no middle ground offered.
The Spark That Lit the Fire
The controversy began with Sashiko, an AI-based code review tool that automatically analyzes kernel patches and emails feedback to authors. Some maintainers raised concerns about the signal-to-noise ratio of AI-generated reviews and the extra workload they create for human reviewers.
Developer Laurent Pinchart pointed to recommendations from the Software Freedom Conservancy on LLM usage in FOSS contributions, suggesting maintainers should “triage and verify” AI reviews before acting on them.
Roman Gushchin pushed back, arguing that requiring human triage before AI can help “makes the point of sashiko, helping maintainers, unachievable.” He noted the SFC document “expresses a very anti-LLM position in general” and asked whether Linux as a project condones that stance.
Torvalds’ response landed like a hammer.
“Linux Is Not One of Those Anti-AI Projects”
The full statement, published on the Linux kernel mailing list, cuts through the noise with characteristic bluntness:
“I realize that some people really dislike AI, but this is an area where I’m willing to absolutely put my foot down as the top-level maintainer.
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.”
This isn’t a nuanced policy proposal. It’s a line in the sand. Torvalds isn’t arguing about whether AI tools have problems, he’s arguing that refusing to engage with them is not an option.
AI as a Tool: The Technical Merit Argument
Torvalds frames AI as a tool like any other, and the kernel community evaluates tools on technical merit, not ideological purity:
“AI is a tool, just like other tools we use. And it’s clearly a useful one. It may not have been that ‘clearly’ even just a year ago, but it’s no longer in question today.”
He draws a direct line to the kernel’s founding philosophy:
“In the kernel community we do open source because it results in better technology, not because of religious reasons. And so we make decisions primarily based on technical merit. Not fear of new tools.”
This is the core of his argument. Torvalds isn’t blind to AI’s flaws, he explicitly acknowledges it can be “a somewhat painful tool” that “keeps finding embarrassing bugs.” But the solution, in his view, isn’t to bury your head in the sand. It’s to make LLM tools help maintainers rather than burden them.
The Social Warrior Dig
The most provocative line in the entire statement is the one that’s generating the most heat:
“This is NOT some kind of ‘social warrior’ project, never has been, and never will be. Sure, the social angle of working on open source is important and often a very motivating part of the project, but in the end that’s a side benefit, not the point of the project.”
This cuts to the heart of a deeper tension in open source. There are developers who see open source as a social movement with ethical obligations, including how tools are built and who benefits. Torvalds is explicitly rejecting that framing. The kernel exists to produce good technology. Everything else is secondary.
Critics on GamingOnLinux were quick to call this “blinkered” and “naive”, with one commenter pointing out that “just a tool” framing ignores the massive theft, resource waste, and economic distortion that generative AI enables. Another noted that Torvalds himself has taken moral stands before, he famously kicked Russian maintainers out of the kernel following the invasion of Ukraine, contradicting the “technology over social issues” stance.
The Community Reacts
The developer community response on Reddit has been largely supportive, with the top comment (741 upvotes) capturing the general sentiment: “You can use AI on Linux submissions, but god help your soul if you submit slop.”
The thread highlights a key distinction that’s often lost in the AI debate: using AI as a tool is different from abdicating responsibility. As one commenter put it:
“They have a great policy, yes, you can use AI tools, but you are completely personally responsible for the code you push.”
Several commenters drew parallels to past technology debates:
“My old-timer brains remember the whole same discussion around IDEs in the late 90s. People insisting that if you use an IDE, it’s because you don’t know how to program for real. I swear it just goes in circles.”
The comparison is apt. Every generation of developers has faced moral panic over new tools, from assembly to C to Java to JavaScript frameworks. AI is just the latest on a long list.
But others pushed back against the “just a tool” framing:
“The ‘just a tool’ framing is willfully ignorant. Hammers, rakes, tape measures, and AI are all the same uncontroversial category.”
The core objection is that generative AI has unique properties, its training data was scraped without consent, its operation requires massive energy consumption, and its output can be indistinguishable from human work, making quality control exponentially harder.
Practical Implications for Kernel Contributions
What does this mean for developers submitting patches? The practical guidance is straightforward:
- AI is permitted. You can use LLMs to write, review, or debug kernel code.
- You are responsible for the output. If your AI-generated code is slop, it will be rejected. The same standard that applies to human-written code applies to AI-assisted code.
- Don’t attack other contributors for using AI. Torvalds has made it clear he will “very loudly ignore” people who argue against others using AI tools.
- The bar for quality hasn’t changed. Good code is good code, regardless of how it was produced.
One Reddit commenter summed it up: “Using a tool poorly is worthy of shade. Using it well is indistinguishable from humans doing it well.”
The Slop Problem
The elephant in the room is that AI makes it trivially easy to generate vast quantities of plausible-but-wrong code. The kernel already has sashiko-bot responding automatically to low-effort AI-generated submissions.
Torvalds’ frustration with AI slop echoes his broader frustration with bad code. He’s not defending slop, he’s defending the tool. The distinction matters. If a human writes bad code, you don’t ban humans. If an AI generates bad code, the response should be to use the AI better, not to ban AI.
This connects directly to Linus Torvalds’ frustration with AI slop in kernel documentation, where he argued that “the AI slop issue is NOT going to be solved with documentation.” The solution, he believes, is technical, not procedural. Better models, better prompting, better review processes, not blanket prohibitions.
Fork or Walk Away
The most striking part of Torvalds’ statement is the invitation to leave. “If somebody has issues with that, they can do the open-source thing and fork it. Or just walk away.”
This isn’t hostility, it’s clarity. Torvalds is signaling that the Linux kernel project is not going to become a battleground for AI culture wars. If you want to fight about the ethics of AI training data or the environmental impact of data centers, do it somewhere else. The kernel project is about making the best possible operating system.
For developers who see AI as an existential threat to software quality, this is a cold dose of reality. The project they love is moving in a direction they can’t accept. Their options are to adapt, to fork, or to find another project.
For developers who see AI as a productivity multiplier, this is validation. They can use the tools they find helpful without being attacked for it.
The Bigger Picture
Torvalds’ statement reflects a fundamental tension in open source: Is it a technical project with social benefits, or a social movement that produces technology? His answer is clear, but it’s not the only answer.
The kernel’s size and influence mean it will set precedents that ripple through the entire open source ecosystem. If the kernel accepts AI tools, corporate projects will feel more comfortable doing the same. If the kernel rejected them, it would have given ammunition to anti-AI voices everywhere.
By choosing technical merit as his north star, Torvalds is betting that AI tools will continue to improve and that the kernel community can build the infrastructure to handle them responsibly. It’s a bet that aligns with his track record of embracing disruptive technologies and making them work at scale.
Whether the bet pays off depends on whether the community can collectively raise its AI game, using models that produce better output, developing review processes that catch AI-specific failure modes, and maintaining the human judgment that distinguishes good code from plausible slop.
One thing is certain: the discussion isn’t over. It’s just getting started.