Artificial intelligence has reached one of the most conservative corners of software development, and the dispute has swiftly escalated into open conflict. Linux creator Linus Torvalds fired back sharply at project contributors who demand limits on automated code review. His suggestion to the dissenters was blunt: create your own kernel branch, or leave the project.
The Spark: Sashiko’s Machine-Generated Reviews
The trigger was the work of Sashiko, an AI system that vets patches before they enter the Linux kernel. Longtime contributor Laurent Pinchart argued that such reviews should not be forwarded to code authors automatically. In his view, developers deserve the right to opt out of receiving machine-generated feedback.
Pinchart’s position aligns with recommendations from the Software Freedom Conservancy. In June, the organization urged open source projects to respect developers who decline tools built on large language models. Nevertheless, a portion of the Linux community found this approach impractical.
The Automation Argument
Developer Roman Gushchin observed that mandatory human review of every machine-generated comment would nullify the chief benefit of automation. If a maintainer must still read, verify, and forward each remark, the system saves no time at all.
Why the Skeptics Worry
The fears of Sashiko’s opponents did not arise from nothing. Developers of the Linux multimedia subsystem once allowed the system to post reviews directly to the main mailing list. The experiment went poorly. The AI invented nonexistent problems, confused contributors with one another, and forced maintainers to spend extra effort untangling its erroneous messages.
Torvalds Draws His Line
Torvalds, however, rejected the idea of walling Linux off from such tools entirely. He argued that artificial intelligence helps uncover bugs, and that human reviews are hardly flawless either. The Linux creator added that he will not allow the project to become an arena for ideological battles over new technologies.
Even so, Torvalds has criticized the mindless use of neural networks before. In late 2025, he called artificial intelligence both a bubble and a revolution at once. In his assessment, such systems excel at bootstrapping new projects, yet perform noticeably worse at the long-term maintenance of complex code.
Earlier Battles Over AI Noise
Earlier in 2026, Torvalds also spoke out against the flood of identical, poorly verified bug reports generated by language models. Following that intervention, such submissions stopped being treated as confidential by default. Now, authors must first validate their findings so that machine-written text no longer overwhelms the closed security channels.
Support Our Threat Intelligence
If you find our technology report and cybersecurity news helpful, consider supporting our work.
Buy Me a Coffee
PayPal
