Linus Torvalds Tells AI Critics to Fork Linux or Walk Away as Kernel Embraces AI Tools

Linus Torvalds Defends AI Use in Linux Kernel Development

Linus Torvalds has delivered a blunt message to critics of AI in Linux: fork the project or walk away. Speaking on a mailing list thread in July about anti-AI sentiment, Torvalds made clear that Linux is not an anti-AI project2

2

. His stance represents a significant shift from October 2024, when he said his approach to AI was to "basically ignore it"2

2

. The Linux creator emphasized that while no one can be forced to use AI, he will loudly ignore anyone attempting to police others' tool choices. This evolution in perspective comes as AI-assisted tools have become increasingly prevalent in Linux kernel development, with Torvalds himself using AI for tasks outside his core expertise, such as redesigning a guitar pedal interface in a language he doesn't typically work with.

Kernel Maintainers Drop AI Disclosure Requirements

At the Linux Kernel Maintainers Summit in Prague, kernel developer Sasha Levin announced that maintainers had decided to deprecate the Assisted-by tag1

1

. Contributors no longer need to disclose AI use in code submissions. This decision sparked controversy, with critics like Roy Schestowitz calling it "AI slop code" and some commenters suggesting Torvalds had flip-flopped under pressure from the Linux Foundation1

1

. However, the policy change doesn't mean low-quality code will pass muster. Torvalds stressed that competent people should use AI as a tool, writing that the "clueless people using AI" problem is different from skilled developers leveraging AI assistance1

1

. When one developer wanted to mandate AI disclosure tags, Torvalds responded sarcastically by suggesting they also stop using compilers and text editors, arguing that understanding submissions matters more than the tools used to create them2

2

.

AI Quality in Bug Reports Improves Dramatically

Greg Kroah-Hartman, a veteran Linux kernel maintainer, reported a dramatic shift in AI-generated bug reports. In March 2026, he noted that months earlier the kernel team received what they called "AI slop"—obviously wrong or low-quality security reports1

1

. But something changed about a month before his statement, and "the world switched" to real, high-quality reports made with AI assistance1

1

. By August 2026, Kroah-Hartman explained that finding bugs is well-suited to AI because AI is a pattern matcher and code consists of patterns2

2

. This improvement in AI-assisted code review has contributed to what Torvalds called the "new normal" of unusually large Linux kernel test releases, with many extra fixes coming from AI-assisted analysis2

2

.

Maintainers Face Mounting Pressure From AI-Generated Patches

While AI has proven valuable for Linux kernel development, it's creating significant strain on maintainers. In March 2026, Torvalds complained that a release candidate came in too large because AI code review resulted in several patch series arriving late in the release cycle when focus should have been on regressions and stability2

2

. He said he'd become more hard-nosed about that kind of churn. The security mailing list became almost unmanageable due to AI-found bugs, with multiple people running the same tools and reporting identical problems2

2

. Speaking in Prague in October 2026, Torvalds admitted that AI is stressing maintainers out to the point that it's become a problem, with close to three-quarters of talks at the Maintainers Summit the previous day centered around making AI tools less stressful to manage2

2

. The fundamental imbalance is clear: patches and bug reports are now cheap to produce with AI assistance, but human time is still required to check, validate, and decide which suggestions merit action.

Torvalds Distinguishes Between Competent Developers and Drive-By Contributors

Torvalds drew a sharp line between competent developers using AI as a tool and what he calls drive-by patches from those lacking genuine commitment. He wants to discourage clueless people submitting AI-generated code without understanding it, but not discourage those who genuinely want to learn and become useful members of the kernel community1

1

. At Open Source Summit Europe, he described AI as a "gateway drug" to programming, praising how it allows new programmers to accomplish things they'd otherwise struggle with1

1

. His rule is straightforward: the test isn't the tool but whether developers can stand behind their work. This approach mirrors broader industry trends, with AI in programming becoming standard practice regardless of critics' objections. As one observer noted, by 2030 or so, people programming without AI assistance will be about as common as hand-weavers are today1

1

. For Linux kernel development, human oversight remains paramount even as AI-assisted tools reshape how code is written and reviewed.