Linus Torvalds to AI opponents in Linux: "Fork the kernel or move on"

Linus Torvalds has decisively ended the debate on AI in Linux kernel development with blunt candor: those who refuse AI-assisted contributions are free to fork. The divide between pragmatism and ethical principles is widening in the open-source community.

Linus Torvalds' Stance

Linus Torvalds has sent a blunt message to developers opposing the use of AI in the development of the Linux kernel. The crux of the message, as reported, is unambiguous: opponents can fork the kernel if they are unhappy with the direction taken, or simply go elsewhere. Torvalds does not want an endless ideological debate about the tools used by contributors.

The Context: A Growing Divide

Torvalds' stance is part of a debate that has persisted for several development cycles of the Linux kernel. The situation is complex because positions are not monolithic:

Pragmatic Side (Torvalds): What matters is the quality of the patch submitted, not the tool used to produce it. If a patch generated with the help of an LLM is correct, clean, well-tested—it deserves to be integrated. AI is just another tool.

Principles Side (Greg Kroah-Hartman and others): Kroah-Hartman’s staging tree has already established a policy of systematically rejecting patches clearly generated by AI without serious review. The issue is not AI itself, but the volume of superficial and poorly reviewed patches that inundate maintainers, increasing the human review workload without providing proportional value.

Documented Reality: Linux kernel network subsystems have been “bombarded” with AI-generated patches (Phoronix, 2026-08-07), confirming that Kroah-Hartman’s staging tree policy is not enough to stem the flow in other subsystems.

Why Torvalds' Position Matters

Torvalds is not just anyone in Linux governance—he has the final say on merges into the main branch. His stance is therefore a strong signal: opposition to AI in the kernel will not receive protection from the top.

This does not mean every AI patch will be accepted. Quality standards remain the same. But the legitimacy of the principle “ban AI-assisted contributions” will not be defended by Torvalds.

Parallels with the Past

In Linux kernel history, Torvalds has already settled similar debates with the same frankness:

• The controversy over the integration of BitKeeper (proprietary VCS, before Git)
• Debates over the integration of proprietary code via drivers
• Quarrels over languages (Rust in the kernel)

In all these cases, Torvalds eventually made a decision, and the ecosystem adapted. AI in the kernel will likely follow the same path: a practical balance between openness to new tools and maintaining quality standards, without an ideological crusade in either direction.

Key Takeaways for Contributors

• Torvalds will not support a policy excluding AI tools
• Quality standards do not change—the tool used is neutral
• The real issue remains the human review workload generated by low-value patches, whether AI-generated or produced by inexperienced humans