Linux Kernel Maintainers Move to Drop AI-Assistance Disclosure Tag

Summary

Linux kernel maintainers have reached consensus to deprecate the Assisted-by tag, which documentation had asked contributors to add when using advanced coding tools, including AI systems. Sasha Levin proposed removing the requirement, and the proposal drew more than 50 replies. Supporters argued that contributors do not consistently follow the rule, that the tag can act as free advertising for proprietary LLM companies, and that the project has not explained why it needs the information. Laurent Pinchart opposed removing it without another disclosure mechanism, saying the rule still sets expectations even if some contributors evade it. Linus Torvalds rejected that position, arguing that understanding the code matters more than the tools used to produce it; he compared mandatory tool disclosure to requiring kernel developers to avoid compilers or text editors. Torvalds has also said that he now likes using AI for tasks outside his strengths, including work on a guitar-pedal project, and described vibe coding as a way to make programming more enjoyable and attract newcomers. At the same time, he warned that AI should be used carefully for important work. He said AI tools are finding real kernel security bugs and long-neglected driver flaws, and that AI is improving the code base overall, but the volume of AI-related activity is putting maintainers under enough stress to become a problem. He also acknowledged that automated reviews from Sashiko can create situations where bots review patches produced or submitted by other bots. Other projects have taken stricter positions: System76 bans AI-generated content in its COSMIC desktop, while NetBSD has banned AI-generated code since 2024. The dispute therefore reflects an unresolved tension between disclosure and enforceability, rather than a simple rejection of AI use.