Linux kernel creator Linus Torvalds (리누스 토발즈) described AI as a useful tool that lowers the barriers to entering programming, but said people should be very cautious when using it for important and serious work. IT outlet TechRadar reported on Thursday that Torvalds made the remarks at an Open Source Summit held by the Linux Foundation in Prague, Czech Republic.

Torvalds said AI could help junior developers and teenagers start programming. He said polished results are now so abundant that beginners can easily feel their small attempts are inadequate, but AI can lower the threshold at the first step and help them keep learning and stay interested. He described AI as an interesting and very useful tool and an entry tool for programming.

He was cautious about putting AI straight into important development work. He said users need to know exactly what they are trying to do and what the output should look like to use AI properly. Torvalds said it would be better to try AI first on toy projects or personal projects.

Torvalds again pointed out that vulnerability reports and patches produced by AI are adding to the burden on Linux maintenance. He has said that in mid-May 2026 the Linux mailing list was nearly unmanageable because of useless AI-generated reports. Different people repeatedly found the same problem with the same tools, piling up duplicate reports, and maintainers spent time reclassifying issues that had already been fixed or passing them on to those responsible.

He said AI keeps finding issues even in code that is hardly used anymore. Torvalds said that as a result, maintainers end up handling important and unimportant patches together. He added that overall, this trend is helping improve the codebase.

The problem is maintainers' stress. Torvalds said about three quarters of the discussion at a maintainer summit focused on how to make AI tools, including code generation and review, less burdensome and more useful. He said the response is shifting not toward using less AI but toward using more of it. He said people are using AI-based patch review to counter a flood of patches, and that bots are even talking to bots.