Linus Torvalds opposes restricting which tools Linux kernel contributors use, as maintainers consider dropping the project’s Assisted-by disclosure tag for AI-assisted code. The proposal reflects a practical problem: the tag appears on only about 10% of 18,533 non-merge commits in Linux 7.3 through release candidate 7, according to Linuxiac.
Watch Desk analysis
What happened
The Linux kernel mailing-list discussion began with a proposal from developer Sasha Levin to remove the Assisted-by: LLM tag. Levin’s revised proposal says the system has proved largely ineffective: contributors may not know they should use it, while others leave it out, so maintainers cannot reliably identify AI involvement from the tag alone.
Torvalds argues that developers’ understanding and responsibility for their code matter more than whether they used AI or another tool. The proposal has support from prominent maintainers, including Greg Kroah-Hartman, but Linuxiac reports it has not yet been merged. Existing documentation on generated content would remain, and contributors would still be responsible for their submissions under the Developer Certificate of Origin.
Why it matters
This is a concrete test of what AI transparency can achieve in a large open-source project. A disclosure rule sounds tidy, but a label that misses most cases may offer maintainers little confidence while adding another box for contributors to tick.
Dropping the tag would not mean AI-generated code gets a free pass. It would shift the emphasis from declaring which tool was used to reviewing the contribution and holding its author responsible. That may be more workable, though it gives maintainers less visibility into how code was produced.
Our read
Torvalds has a point about judging the work rather than banning the wrench. But if disclosure is going, review expectations need to be clear enough that “the author is responsible” means something in practice. For now, this is a proposal, not a completed change to kernel policy.
What to watch
- Whether Levin’s revised proposal is merged.
- How the kernel’s contribution documentation and checkpatch.pl change if it is adopted.
- Whether maintainers set out other ways to handle AI-assisted submissions.
Discussion spark: Should open-source projects require contributors to disclose AI assistance even when the label is easy to miss or omit, or is responsibility for the submitted code enough?
Sources and evidence
- Source update (16 October 2026, 10:45 UTC)
Watch Desk is operated by WittyWires as an independent cross-cutting AI news tracker. It does not speak for the organisations or people it covers.