When AI/LLM agents are used in the creation of Linux kernel patches, the policy for a while now has been that it should be specified using an "Assisted-by" tag as part of the patches/commits. But Linux kernel developers this week have been discussing whether to revise that policy or to potentially eliminate it.
Prominent Linux kernel developer Christian Brauner kicked the hornet's nest yesterday in starting a Linux kernel mailing list thread seeking to simplify the attribution of AI/LLM agents. Current Linux kernel documentation notes the Assisted-by tag should contain just not the LLM/name but also the model version and other detailed information. Brauner argued:
He also suggested to perhaps just remove the disclosure need entirely given its lack of usefulness. Short of dropping the disclosure need with the "Assisted-by" tag, he suggested making it simpler with just needing the common name of the AI/LLM being used without any extra details.
In response there have been dozens of messages with other kernel developers sharing their opinions. All the usual common traits have materialized from acknowledging that not everyone is following the "Assisted-by" guidelines, the "free promotion" to proprietary/commercial LLMs, the additional clutter it adds when reading patch messages, etc. There is also the lack of context for how the AI/LLM was used whether it generated the patch itself or just assisted in testing, code review, or other purposes.
Of course, many also in the camp of being firmly against any use of generative AI in open-source development.
Those wanting to go through all the wide-ranging comments can do so via
Beyond that, today a separate
Linus Torvalds has yet to chime in on either thread. We'll see where this discussion leads in refining or removing the Linux kernel AI/LLM attribution guidelines.
Prominent Linux kernel developer Christian Brauner kicked the hornet's nest yesterday in starting a Linux kernel mailing list thread seeking to simplify the attribution of AI/LLM agents. Current Linux kernel documentation notes the Assisted-by tag should contain just not the LLM/name but also the model version and other detailed information. Brauner argued:
"I find it very irritating that our git history has effectively started to function a bit like a free advertising platform for a bunch of AI companies and their proprietary agents and models."
He also suggested to perhaps just remove the disclosure need entirely given its lack of usefulness. Short of dropping the disclosure need with the "Assisted-by" tag, he suggested making it simpler with just needing the common name of the AI/LLM being used without any extra details.
In response there have been dozens of messages with other kernel developers sharing their opinions. All the usual common traits have materialized from acknowledging that not everyone is following the "Assisted-by" guidelines, the "free promotion" to proprietary/commercial LLMs, the additional clutter it adds when reading patch messages, etc. There is also the lack of context for how the AI/LLM was used whether it generated the patch itself or just assisted in testing, code review, or other purposes.
Of course, many also in the camp of being firmly against any use of generative AI in open-source development.
Those wanting to go through all the wide-ranging comments can do so via
this LKML thread.Beyond that, today a separate patchwas proposed by Linux kernel engineer Jeff Layton. He's taking the step of proposing a patch to drop the LLM / Assisted-by requirement from the Linux kernel entirely. His rationale in the patch message:"We've had this requirement in place in the Documentation for several months, but it's becoming clear that the signal to noise ratio from this is quite low.
1/ It's not universally followed. While many people do try to attribute
the LLMs in good faith, not everyone does for various reasons.
2/ It basically serves as free advertising for proprietary LLM companies.
3/ It's not clear why we want to collect this info in the first place.
Given that the data this provides is flawed at best and is being collected for a purpose that isn't clear, let's just kill the requirement for these tags from the kernel at large."
Linus Torvalds has yet to chime in on either thread. We'll see where this discussion leads in refining or removing the Linux kernel AI/LLM attribution guidelines.