Re: [MAINTAINERS SUMMIT] The role of AI and LLMs in the kernel process

[James Bottomley <James.Bottomley@HansenPartnership.com>]

On Mon, 2025-12-08 at 08:00 +0100, Jiri Kosina wrote:
> On Sun, 7 Dec 2025, Sasha Levin wrote:
> 
> > 3. Legal risk tolerance:
> > 
> > DCO clause (a) requires certifying "I have the right to submit it
> > under the open source license." With AI training data provenance
> > unclear and litigation ongoing, how cautious should we be? Some
> > advocate waiting for legal clarity; others argue the legal concerns
> > are overblown and we should focus on practical guardrails.
> 
> In either case, isn't all the legal responsibility with the
> submitter, who submitted under the terms of DCO, and thus this
> particular aspect is not really a concern for us (maintainers) on the
> receiving side as per the (c) DCO clause?

In terms of liability for maintainers, yes, you're right.  However, in
terms of legal risk for the kernel itself, no, because the problem is
the situation where we've accepted loads of AI code that then has to be
reverted if a court decides AI code is a derivative work of its model.
I think this latter is where we need a collective agreement to risk
tolerance.

However, you can put me down as thinking the risk to the kernel of an
adverse court decision on AI is acceptably small.

Regards,

James