Torvalds says Linux will not reject AI tools

Linus Torvalds has told Linux kernel contributors that the project will not define itself as anti-AI, drawing a line in a debate over whether AI-assisted work belongs in one of the technology industry’s core infrastructure projects. No money, valuation or commercial launch is involved, but the position matters because Linux maintainers set norms for a development model used across enterprise software, cloud infrastructure and open source.

In a post on the Linux kernel mailing list archive lore.kernel.org, Torvalds responded to criticism of AI use in kernel development by saying he was prepared to intervene as the project’s top-level maintainer. Linux, he wrote, is not an anti-AI project, and people who object can use the standard open source remedy: fork the codebase. He also said contributors can leave.

Torvalds described AI as another tool available to developers and maintainers, not a principle that should be rejected on ideological grounds. He said its usefulness is now clear, while allowing that AI can create work for maintainers and can be uncomfortable when it identifies bugs that humans missed.

The practical issue for the kernel project is not whether AI exists, but whether AI-assisted bug reports, reviews and patches save time or create low-quality work that maintainers must clean up. Torvalds said the answer should be to make large language model tools useful to maintainers rather than prohibit others from using them. He added that no one is being required to use AI.

His comments mark a more permissive public stance than the one he took in October 2024, when he characterized most AI activity as marketing noise and said he preferred to ignore it for the time being. At that point, he said the field might look different over a five-year period. The latest mailing list post suggests he now sees enough practical value to reject blanket opposition inside the kernel community.

Torvalds also framed the issue as part of a broader view of open source. He said the kernel project uses open source methods because they produce better technology, not for religious reasons. Social collaboration, in his telling, is useful, but the project’s decisions should be based primarily on technical merit rather than anxiety about new tools.

That position lines up with comments from senior Linux maintainer Greg Kroah-Hartman earlier this year. In a March interview with The Register’s Steven J. Vaughan-Nichols, Kroah-Hartman said AI-assisted bug reports and code review had recently become much more useful. He said open source projects were starting to receive AI-generated reports that were real and good, after a period in which many such reports were not helpful.

The concern has not gone away. Some open source maintainers have complained that AI-generated bug reports add to burnout when they are shallow, wrong or duplicative. Others have raised concerns about code produced through so-called vibe coding, where developers rely heavily on generative systems without enough understanding of the result.

Torvalds acknowledged that AI is imperfect. His argument was that human work is also imperfect, and that pointing to AI’s defects is not enough to justify blocking it. For the kernel, the test he is setting is operational: whether a tool reduces maintainer burden and improves the code, rather than whether it fits a broader cultural position on AI.

This story draws on original reporting from The Register.