Linux Maintainers Can Now Submit AI-Generated Code Without Declaring It

Linux maintainers have voted to no longer require disclosure of AI-generated code submissions, as announced at the Maintainers Summit held on October 8, 2026, in Prague, Czechia. The Linux kernel's Assisted-by tag, which previously mandated the declaration of LLM usage, will no longer be necessary for submissions. Linux's creator, Linus Torvalds, expressed his disapproval of such a ban, describing it as "utterly stupid," comparing it to removing text editors that help ensure the correctness of commit messages.

Torvalds stated that he expects kernel contributors to be intelligent and comfortable using new and innovative tools.

During the summit, Torvalds used AI, specifically Anthropic's Claude Code, to assist in creating a patch that aimed to solve an issue related to cache contention when multiple CPUs access the same file. The development branch, which Torvalds posted, claims to have improved performance by a factor of 23 in tests conducted using the Claude Code assistant. However, he acknowledged that real-world impact would likely be less significant, as it would not represent a typical load scenario.

Despite previously criticizing the notion that AI could write up to 99% of code, Torvalds submitted a 10-patch series, which included the AI-assisted solution. Reviewers identified a race condition and raised concerns about memory reclamation in the patch. Torvalds handed the development of the patch to Kiryl Shutsemau, a Linux kernel developer, who had previously worked on optimizing the same bottleneck.