By Hamza Faisal

Linus Torvalds has never been one to mince words, and this week the Linux creator aimed his trademark bluntness at the growing trend of AI tool restrictions — calling bans on AI-generated code "utterly stupid" as the Linux kernel moves to scrap the rule that forced developers to disclose AI assistance in their patches.

The decision crystallized at the Linux Maintainers Summit in Prague on October 8, 2026, where attendees reached consensus to deprecate the "Assisted-by" tag — a commit trailer introduced earlier this year to identify patches developed with the help of AI coding assistants. A proposal from kernel developer Sasha Levin would now remove the requirement from the kernel's contribution documentation and strip its handling from the checkpatch.pl review script.

The tag that told maintainers almost nothing

Levin's case for removing the tag rests on cold numbers. Of the 18,533 non-merge commits in Linux 7.3 through release candidate 7, only about 10 percent carried an Assisted-by tag — even though AI tools are believed to be in use far more widely. Many contributors never knew the requirement existed; others deliberately skipped it. The result: maintainers cannot reliably tell whether AI was involved simply by looking for the tag.

Notably, Greg Kroah-Hartman — the kernel's second-in-command, who had previously supported keeping the tag — has now swung behind its removal.

"Don't use compilers for any of your kernel submissions"

The discussion turned combative when developer Laurent Pinchart argued that clear expectations around AI use still have value, even if some contributors ignore them. "I really don't get the unenforceable argument here," Pinchart wrote.

Torvalds's reply was instant and scathing: "I don't understand this whole idiotic concept of 'don't use AI for X.'" Then, in classic fashion, he turned the logic on its author — developers, he suggested, should be forbidden from using compilers for their kernel submissions and required to write assembly to prove they truly understand what their code does, and should not use editors to write commit messages either. "Do you not see how UTTERLY STUPID those kinds of rules would be?" he wrote. "We're not going to make stupid rules that say 'you can't use tools for X.'"

His stated standard is simpler: "The kind of expectations I have is that people do their best and I expect people to understand what they are doing. But that understanding has very little to do with what tools they use."

Linux goes its own way on AI

The decision puts Linux sharply at odds with projects that have moved to ban AI-assisted contributions outright — including Godot and RPCS3, both of which prohibited AI pull requests this year.

To be clear, the kernel is not lowering its bar. AI-assisted patches are still accepted, but only when a human reviews them, adds the standard sign-off, and keeps the code compatible with the GPL-2.0-only license. Unresolved concerns remain: reviewers worry about subtle errors slipping through unnoticed, about accountability when nobody can say who "wrote" a patch, and about agents reproducing copyrighted code.

For now, though, the message from the top is unambiguous. Torvalds himself has said Linux is not an anti-AI project — and the kernel's future will increasingly be built with AI tools, not without them.

By Hamza Faisal