Linus Torvalds has issued a definitive and characteristically blunt defense of AI tooling within the Linux kernel development process, directly challenging detractors and setting a clear policy for the project’s future. In a strongly worded email to the Linux kernel mailing list, Torvalds declared that the kernel project is not and will not become “one of those anti-AI projects,” and he is prepared to use his authority as the top-level maintainer to enforce this stance. “I will absolutely put my foot down as the top-level maintainer,” Torvalds wrote. Anyone who disagrees with this direction “can do the open-source thing and fork it. Or just walk away.”
Torvalds Defends AI-Assisted Code Review Against Backlash
The exchange was prompted by developer Roman Gushchin, who raised concerns about what he described as a pervasive “anti-LLM position” within the kernel community. This attitude, Gushchin argued, was undermining the objectives of Sashiko, an AI-powered code review system developed under the Linux Foundation. Sashiko is designed to automate the review of proposed patches by using prompts tailored specifically to the Linux kernel and its own communication protocol. It can import patches directly from mailing lists or local Git repositories and operates with various large language model providers.
Torvalds acknowledged that a segment of the community holds anti-LLM views but made it unequivocally clear that this sentiment does not reflect the project’s official position. He argued that AI is “clearly a useful one,” even if its value was less apparent a year ago. “Anybody who doubts that clearly hasn’t actually used it,” Torvalds wrote, dismissing efforts to restrict the use of AI tools as a form of willful ignorance.
Technical Merit Over Ideology: Torvalds Rejects Anti-LLM Stance
Torvalds did not shy away from the potential downsides of integrating AI into the development workflow. He acknowledged that AI can be “painful,” citing scenarios where it creates additional work for maintainers or uncovers “embarrassing bugs.” However, he argued that these problems are not a reason to reject the technology outright. The solution, in his view, is not to “put your head in the sand and sing ‘La La La, I can’t hear you’ at the top of your voice.” He drew a parallel to human intelligence, noting that it too is far from perfect.
The core of Torvalds’ argument rests on a principle of technical meritocracy. “Not fear of new tools,” he stated. The kernel project makes its decisions based on the technical merits of the technology in question, not on ideological opposition. AI, like any other tool, is subject to this same evaluation, and no one is forced to use it. However, Torvalds made it clear that he will “very loudly ignore” people who attempt to dissuade others from adopting AI tools. A dedicated guide for kernel maintainers on working with Sashiko is available on GitHub, signaling the project’s commitment to integrating the tool into the official workflow.
What Does This Mean for the Future of Linux Development?
Torvalds’ directive represents a significant moment for the open-source community. It signals that one of the most influential figures in software development sees AI not as a passing trend or a source of low-quality output, but as a legitimate and valuable tool for maintaining and evolving a massive, complex codebase. For developers, maintainers, and contributors, the message is clear: resistance to AI-assisted development is not a viable stance within the Linux kernel project. The emphasis on technical merit over ideological purity is a hallmark of Torvalds’ leadership, and this latest statement extends that principle into the era of AI.
Who Should Try This Now
For anyone involved in Linux kernel development, the immediate implication is that tools like Sashiko are not only welcome but actively endorsed at the highest level. Developers curious about integrating AI into their workflow for patch review or code analysis should consider exploring Sashiko’s capabilities and the accompanying maintainer guide. The most practical step is to review the Sashiko repository on GitHub and experiment with importing a patch from a mailing list or local repository to evaluate how an LLM-based review system might augment, rather than replace, the human review process.
