Open source is not fighting about AI. It appears to be fighting about where in the pipeline to draw the line.

The AI fight tearing through open source is not about the tools. It appears to be about which step in the pipeline gets policed.

Linux just answered that question for its own project. The flashpoint is Sashiko, an opt-in code review tool that comments on proposed kernel patches and claims to find 53.6% of bugs on patches that already cleared human review, per Tom's Hardware's reporting of the Sashiko project page. That number is the load-bearing detail. Once a tool catches more than half the residual bugs after humans have looked, the live question is no longer whether to use it. It is whether to filter its output before it lands on a contributor.

Laurent Pinchart wants the filter. Linus Torvalds, who dismissed AI as overhyped in 2024, does not. His reply on the Linux Kernel Mailing List, covered by Phoronix and The Register, was blunt: Linux is not an anti-AI project, and critics can fork it or walk away.

The portable mechanism: Gentoo, Curl, and Ghostty all drew their AI lines on the author side, restricting AI-generated submissions. Linux drew its on the reviewer side, allowing AI comments and rejecting the triage step. Both moves are about locating the rule in the pipeline. Projects that frame the fight as ideology are skipping the only question that decides outcomes.