Linus Torvalds Backs AI in Linux Kernel Development But Warns Against Low-Quality Drive-By Patches

Linus Torvalds Declares AI Tools Welcome for Competent Developers

Linus Torvalds has made his position clear: AI tools belong in the programming toolkit, and the Linux kernel will not ban their use. At the Linux Kernel Maintainers Summit in Prague, kernel developer Sasha Levin announced that maintainers decided to deprecate the Assisted-by Tag, meaning contributors no longer need to declare AI use in code submissions1

1

. This decision signals a pragmatic shift in how the open-source community approaches AI-generated code.

Torvalds himself has used AI to locate bugs, write fixes, and even generate commit messages for the Linux kernel2

2

. When questioned about setting expectations around AI usage, he responded forcefully, calling the idea of banning AI tools "utterly stupid" and comparing it to refusing compilers or text editors3

3

. His message to critics: fork the project or walk away.

Quality Standards Trump Tool Choice in Code Submissions

While Torvalds supports AI tools for coding, he's established a non-negotiable standard: competence matters more than methodology. "I really think we want competent people to use AI as a tool," Torvalds stated, emphasizing that the focus should remain on the developer's understanding rather than their toolkit1

1

.

The distinction lies in intent and capability. Torvalds welcomes developers who use LLMs to learn and become valuable members of the open-source community. What he opposes are drive-by patches from people who generate code without understanding it, seeking recognition without commitment. "We all start clueless, but some people spend the time and effort to become something more," he explained1

1

.

This balanced approach means AI use in code submissions is permitted, but human oversight and understanding remain essential. The rule is simple: if the code is bad, it doesn't make it in, regardless of whether it was written by humans or generated by AI2

2

.

AI Bug Reports Shift From Slop to Valuable Contributions

The quality of AI-generated contributions has improved dramatically. Greg Kroah-Hartman, a veteran Linux kernel maintainer, noted that months ago the community received "AI slop"—obviously wrong or low-quality security reports. But something changed about a month before March 2026. "Now we have real reports," Kroah-Hartman confirmed, adding that AI-generated bug reports across all open source projects have become genuinely useful1

1

.

By August 2026, Kroah-Hartman described bug finding as a job suited to AI because it excels at pattern matching, and code is fundamentally patterns3

3

. Torvalds himself attributed unusually large test releases to AI-assisted code review, calling the influx of fixes "the new normal"3

3

.

Vibe Coding Has Its Place But Not in Production Systems

Torvalds has defended vibe coding—using AI to quickly generate functional code—as a gateway to programming. At Open Source Summit Europe, he called it "a wonderful way to find joy in programming" that allows new programmers to accomplish tasks that would otherwise prove difficult1

1

. He even stated he "loves the concept of AI as a gateway drug" to programming1

1

.

However, Torvalds draws a clear line between personal projects and kernel contributions. Vibe coding a fun project is acceptable; submitting it to the Linux kernel without proper understanding is not2

2

. He personally uses AI for tasks outside his expertise, admitting, "I don't do Java, I do kernels," when describing how he used AI to redesign a guitar pedal interface3

3

.

Maintainers Face Growing Pressure From AI-Generated Submissions

The proliferation of AI tools has created an imbalance that's stressing maintainers. While patches and bug reports have become cheap to produce, human time is still required to review, validate, and decide which suggestions merit action3

3

.

In March 2026, Torvalds complained that the security mailing list had become almost unmanageable because multiple people running the same AI tools were reporting identical bugs3

3

. A release candidate arrived too large because AI-assisted code review generated numerous patch series when the focus needed to be on regressions and stability3

3

.

By October 2026 in Prague, Torvalds admitted AI is stressing maintainers to the point where it's become a problem. Close to three-quarters of talks at the Maintainers Summit centered on making AI tools less burdensome3

3

. Watch for potential policy adjustments as maintainers develop strategies to manage the volume without sacrificing quality or burning out reviewers.