Linus Torvalds: AI Makes Programming More Fun, but Kernel Work Needs Caution

At Open Source Summit Europe in Prague in October 2026, Linux creator Linus Torvalds discussed his views on AI in programming and Linux kernel development with Dirk Hohndel, head of Ericsson Software Technology. Torvalds said he really likes using AI, finding it makes programming much more enjoyable and especially helpful for newcomers. However, he stressed that AI must be used carefully in real and important work such as the Linux kernel, where AI-generated patches and bug reports can overwhelm maintainers.

Torvalds recalled starting programming around 1981 and noted that the bar in software engineering has risen dramatically. He described AI as a "gateway drug" that lets new programmers do things they would otherwise struggle with. He gave an example of using AI to write Java code for his guitar pedal hobby project, admitting the first attempt was poor but that he could guide it using his own C implementation.

On kernel development, Torvalds mentioned Sashiko, an agentic code review system whose public reviews now appear on the Linux Kernel Mailing List (LKML). Some subsystem maintainers now expect patches to have received such a review before acceptance. He acknowledged AI tools can identify genuine security problems and surface long-ignored issues in old drivers, improving the code base, but also stress maintainers. He cited concerns about convincing but fabricated bug reports and narrow patches that act as "mindless band-aid" fixes.

Torvalds said about three-quarters of the preceding day's Linux Kernel Maintainer Summit discussion concerned making AI generation and review less stressful and more useful. The bottleneck is not getting patches but getting useful work through review without exhausting maintainers. He concluded that AI code checking is here to stay.