Linus Torvalds: Linux Is Not Anti-AI — Technical Merit Is the Only Standard

Linus Torvalds Takes a Clear Stance

Linux kernel creator and top maintainer Linus Torvalds recently weighed in on the ongoing debate within the community over AI tools, making his position unmistakably clear: Linux is neither an "anti-AI" project nor a "social warrior" project. His remarks quickly sparked widespread discussion on Reddit and other tech communities.

Torvalds, never one to mince words, put it bluntly: "I realize some people really hate AI, but on this issue, as the top maintainer, I'm willing to take an absolute stance. Linux is not the kind of project that is anti-AI, and if anyone has a problem with that, they can do what open source is meant to allow — fork it, or just walk away."

This "take it or fork it" framing is classic open-source logic for resolving disagreements, and it once again underscores Torvalds's firm grip on the project's direction. Fork is a core right granted to every user under open-source licenses: anyone can obtain the source code, create an independent branch under the license terms, and develop it however they see fit. Famous historical forks include LibreOffice from OpenOffice.org and MariaDB from MySQL. By invoking this logic, Torvalds is essentially saying: freedom comes from the forkability of code, not from a veto over the project's direction.

Since Torvalds released Linux in 1991, it has evolved into one of the largest and most widely contributed-to open-source projects in the world. The kernel codebase now exceeds 30 million lines, and each release cycle (roughly 9–10 weeks) merges tens of thousands of patches from hundreds of companies and individual developers worldwide. As a result, the top maintainer's stance carries enormous influence across the entire ecosystem — this statement's impact extends far beyond a single community dispute.

AI Is a Tool, and Its Value Is No Longer Up for Debate

Torvalds's core argument is that AI is fundamentally just a tool — no different in nature from any other tool used in kernel development — and it is clearly useful.

He highlighted a shift in perception: "Even a year ago, this might not have been so 'obvious,' but today it's beyond question. There are still other things worth discussing around AI (like where its business model ultimately leads), but whether it's useful is no longer one of them. Anyone still questioning that clearly hasn't actually used it."

This judgment reflects the rapid penetration of generative AI — and large language models (LLMs) in particular — into software engineering. Large Language Models (LLMs) are deep learning models based on the Transformer architecture, pretrained on massive text corpora. Representative products include GitHub Copilot (based on OpenAI Codex/GPT series) and Google Gemini Code Assist. In software engineering contexts, LLMs are now widely used for code completion, code review assistance, automated unit test generation, vulnerability scanning, and converting natural language requirements into code scaffolding. According to survey data released by GitHub in 2024, developers using Copilot saw an average coding speed improvement of approximately 55% — a productivity gain of this magnitude has effectively closed the empirical debate over whether AI is useful. For a relentlessly pragmatic engineer like Torvalds, there has always been only one criterion: does it solve real problems?

Confronting AI's "Pain Points": Improve, Don't Retreat

Interestingly, Torvalds didn't shy away from acknowledging AI's downsides. He admitted that AI "can also be a pretty annoying tool" — both from the standpoint of maintainer workload and from the angle of "it keeps turning up embarrassing bugs."

But he emphasized that the right response is not to "bury your head in the sand and pretend you can't hear it." The real solution is: "making sure these LLM tools are helping maintainers rather than just creating more work for them."

This is a constructively framed position. The Linux kernel's patch submission process relies heavily on human review: contributors send patches to mailing lists, which are reviewed layer by layer by subsystem maintainers before being consolidated by Torvalds. When AI tools are used to automatically generate patches or vulnerability reports at scale, the review burden on maintainers can grow exponentially. In 2024, the University of Minnesota was temporarily banned from submitting to the kernel after researchers used automated methods to submit patches with intentional defects as part of a security study — an incident that vividly exposed the tension between AI-assisted submissions and the kernel's trust-based review model. Torvalds acknowledges that AI introduces new pressure on the kernel maintenance workflow, but argues the answer lies in improving how tools are integrated and how workflows are designed — not in simply rejecting the technology.

He also added: "We don't force anyone to use it, but I will very loudly ignore anyone who tries to stop others from using it." This draws a clear line: using AI is a personal choice; preventing others from using it is not acceptable.

"Natural Intelligence" Isn't Perfect Either

When faced with criticism that AI makes mistakes and isn't reliable enough, Torvalds offered a response with more than a hint of irony: "AI isn't perfect. But good lord, anyone pointing out AI's flaws had better also look in the mirror and think about themselves. Because natural intelligence isn't always that reliable either."

This analogy cuts to the heart of much AI criticism — people tend to hold AI to a near-perfect standard while treating human errors as a given. Coming from a kernel maintainer who processes enormous volumes of code submissions daily and knows all too well how ubiquitous human-introduced bugs are, this is a particularly grounded observation.

The Nature of the Kernel Project: Technology First

Toward the end of his statement, Torvalds returned to the core principle he has always upheld — the Linux kernel project is about, and will always be about, technology itself.

He said: "Of course, the social aspects of participating in open source matter and are often a very motivating part of the project, but at the end of the day that's a side effect, not the point. This is absolutely not some kind of 'social warrior' project — it never has been and never will be."

He went on to articulate his open-source philosophy: "In the kernel community, we do open source because it produces better technology, not out of some quasi-religious belief. So we make decisions primarily based on technical merit, not out of fear of new tools."

At a time when the open-source world is becoming increasingly politicized, Torvalds's insistence on technical merit as the sole decision-making criterion stands out sharply. In recent years, debates within the open-source community over Codes of Conduct, contributor diversity policies, and the ethical boundaries of technical decisions have grown increasingly heated. The Linux kernel community's official adoption of the Contributor Covenant in 2019 triggered a strong backlash from some community members at the time. Since then, discussions around contributions from Russian developers and the copyright status of AI-generated code have continued to challenge the boundary between "technical standards" and "social/political standards." Torvalds's reaffirmation of a "technology-first" stance is, in this context, a renewed confirmation of the project's governance philosophy.

Conclusion: Pragmatic Technical Leadership

The reason Torvalds's statement resonated so widely is that it reflects a rare pragmatism. At a moment when AI is triggering global anxiety — with resistance and hype existing side by side — the technical leader of the world's most important open-source project chose to cut through the noise with a clear verdict: AI is a tool. If it's useful, use it. If it's imperfect, improve it. Fear gets you nowhere.

For the broader open-source ecosystem, Linux's stance serves as a bellwether — AI-assisted code contribution, review, and maintenance will receive formal recognition in the world's most consequential open-source project. At the same time, Torvalds has offered a reference model for how to think rationally about AI: neither blind worship nor throwing the baby out with the bathwater. Everything comes back to technical value itself.

Key Takeaways