Linux Kernel May Drop Its "Assisted-by" AI Tag as Torvalds Rejects Bans on AI Tools
Earlier this year the Linux kernel added an Assisted-by: tag so contributors could mark patches developed with help from AI coding tools. Now Sasha Levin has proposed removing it again, and the discussion drew in Linus Torvalds.
Why remove it
Levin’s revised patch argues the tag has not worked. Of the 18,533 non-merge commits in Linux 7.3 up to rc7, only about 10% carried it, while AI use is believed to be far more widespread. Many contributors do not know about the rule, and some leave the tag out on purpose. A requirement that is mostly ignored gives maintainers a misleading picture.
The patch would drop the tag requirement from the contribution documentation and remove its handling from checkpatch.pl. Greg Kroah-Hartman, who previously supported the tag, now supports removing it.
What stays
Removing the tag does not loosen responsibility. The existing documentation on generated content remains, and contributors still answer for everything they submit, including the Developer Certificate of Origin sign-off that the code can legally be contributed.
Torvalds’ view
Torvalds opposed restricting AI tools in kernel development. His argument: what matters is whether the developer understands the code they send, not which tools helped produce it. He mocked the idea of tool bans by suggesting developers might as well be banned from compilers and made to write assembly.
Not everyone agrees. Laurent Pinchart argued that clear expectations about AI use are still needed even if some people ignore them.
The revised patch has support from several senior maintainers but has not been merged. Other projects are going the opposite way: COSMIC recently stopped accepting LLM-generated pull requests, and Flathub just replaced its AI ban with mandatory disclosure.