On 13 October 2024, Linus Torvalds blasted the Linux kernel’s AI‑disclosure policy as “utterly stupid,” prompting kernel maintainers to abandon the Assisted‑by tag that had required contributors to mark commits that used advanced coding tools. The decision follows a patch proposal by Sasha Levin and a heated debate among developers about how, or whether, to disclose AI assistance.
Key Takeaways
- Linus Torvalds called AI‑disclosure rules “utterly stupid” in a kernel mailing‑list post.
- Kernel maintainers voted to deprecate the Assisted‑by tag and remove its documentation.
- Debate centers on whether AI tool usage should be publicly disclosed or merely tracked internally.
- Other projects are taking opposite stances: System76 bans AI‑generated code, NetBSD has banned it since 2024.
- AI tools are finding security bugs but also increasing maintainers’ workload.
| Metric | Value | Why it matters |
|---|---|---|
| Date of debate | 13 Oct 2024 | Marks the moment the policy was publicly challenged. |
| Key figure | Linus Torvalds | Primary advocate for the kernel’s direction. |
| Tag removed | Assisted‑by | No longer required in commit messages. |
| Opposing developer | Laurent Pinchart | Voiced concerns about losing disclosure mechanisms. |
The Assisted‑by Tag and Its Purpose
Since January, the kernel’s documentation had asked anyone using “any sort of advanced coding tool” to add the Assisted‑by tag to their commit message. The intent was to surface AI or other automated assistance in the code review process, giving maintainers visibility into potential quality or security implications.
When Sasha Levin proposed a patch on 9 October, the kernel mailing list received more than 50 replies. Levin’s message read: “The Maintainers Summit reached consensus to deprecate the Assisted‑by Tag. Remove the documentation requiring it.” The proposal aimed to streamline the commit workflow and reduce what some developers saw as unnecessary overhead.
The Debate That Sparked the Change
Jeff Layton, a Red Hat kernel developer, offered three reasons to drop the tag. First, not everyone follows the policy. Second, the tag “basically serves as free advertising for proprietary LLM companies.” Third, the kernel has no clear purpose for collecting that information. Layton’s points resonated with many in the community, but not all.
“I’m not happy removing the tag without adding another mechanism for required disclosure,” said Laurent Pinchart. “Without that, I can foresee submitters pushing back on the question.” He added that the rule sets expectations and that the kernel would continue to reject non‑compliant contributions.
Torvalds’ response was swift and scathing. “Everything to do with your hangups,” he wrote. He then issued his own rule: “don’t use compilers for any of your kernel submissions.” He demanded that contributors write in assembler to prove machine‑level understanding and avoid editors that hide typing mistakes. The tone was unmistakably dismissive of the AI‑disclosure debate.
Torvalds’ Take on AI in Kernel Development
Torvalds’ comments came just days after he spoke at the Open Source Summit Europe in Prague. He admitted that his view of AI had shifted: “Now I’m at the point where I actually really like using AI.” He explained that he uses AI for tasks he is “bad at,” such as creating a modern interface on a small screen for a guitar‑pedal project written in C. He described AI as a “gateway drug” for new programmers and praised its ability to find security bugs in the kernel and flaws in drivers untouched for two decades.
Despite the benefits, Torvalds warned that AI “is improving our code base,” but it also “stresses maintainers out to the point that it’s a problem.” He noted that at the Maintainers Summit the day before, about three quarters of the talks focused on making AI tools less stressful. He also referenced Sashiko, an agentic review system whose reviews now appear on the mailing list, and acknowledged that “bots talking to bots” is a real phenomenon.
Industry Reactions and Alternative Policies
Not all projects are embracing AI. System76 has banned AI‑generated content from its COSMIC desktop. NetBSD, which banned AI‑generated code in 2024, has since raised $120,282 in donations against a $50,000 target, according to The Register. These moves illustrate a spectrum of attitudes toward AI in open‑source development, from cautious adoption to outright prohibition.
Why it Matters
The kernel’s decision to drop the Assisted‑by tag signals a broader debate about transparency, quality control, and the role of AI in critical software. For maintainers, the change may reduce administrative overhead but also removes a visible cue that a patch was assisted by an automated tool. For contributors, it could lower barriers to entry but also erode a safety net that flagged potential AI‑related issues. The move also reflects the tension between open‑source ideals of minimal friction and the practical need to guard against subtle bugs introduced by powerful language models.
Frequently Asked Questions
Who made the decision to drop the Assisted‑by tag?
The kernel maintainers reached consensus at the Maintainers Summit, following a patch proposal by Sasha Levin on 9 October.
What was the original purpose of the Assisted‑by tag?
It was intended to require contributors to disclose when they used advanced coding tools, such as AI or other automated assistance, so maintainers could review those patches with additional scrutiny.
Why did Linus Torvalds criticize the AI‑disclosure rules?
He called them “utterly stupid” and argued that the policy imposed unnecessary friction without clear benefits, especially since the kernel had no mechanism to enforce or verify compliance.
What alternative policies are other projects adopting?
System76 bans AI‑generated content from its desktop, while NetBSD banned AI‑generated code in 2024 and has seen significant donations as a result.
How are AI tools affecting kernel development?
AI tools are finding security bugs and long‑unmaintained driver flaws, but they also increase the workload for maintainers and can lead to “bots talking to bots” scenarios.
Bottom Line
The Linux kernel’s removal of the Assisted‑by tag marks a decisive shift away from mandatory AI disclosure in commit messages. While the change streamlines contributions, it also removes a layer of transparency that some developers valued. As AI tools become more capable, the community will need to balance openness, efficiency, and security—an ongoing conversation that will shape the future of kernel development.
Reporting based on: original source.
Comments
Post a Comment