Linux founder Linus Torvalds says AI is a “gateway drug” into programming — but cautions about using AI for “serious” work

(Image credit: Future) For all the doom and gloom headlines AI is driving these days, one important person just gave a refreshing new take on it - none other than Linus Torvalds, the founder and key maintainer of the Linux kernel.At the recent Open Source Summit, organized by the Linux Foundation in Prague, Czechia, Torvalds took the stage to reminisce about the early days of Linux, how he ended up building Git, but also how AI changed what it meant - and how it felt - to be a programmer today.His key takeaway is that AI can make programming less intimidating for junior coders and teens looking to find their way in an otherwise scary, hyper-advanced industry.“People love to hate AI. I think it's an interesting tool, very useful. It's a gateway drug and a way to do really good code,” he said, somewhat jokingly.On AI and “sucking”Speaking more about AI, Torvalds said that in his early days as a programmer, he made computer games which “sucked”, but it was fine because everything else also “sucked” back then.“At the time, the bar was pretty low, right?,” Torvalds said, adding that he didn’t feel ashamed of his work when comparing it to what was already out there. However, younger coders walking into the industry today can easily feel ashamed of their “pitiful, small efforts”, when comparing themselves to all of the polished, professional work around them.And that is exactly where AI can help, he feels. It can lower the bar to entry, so that those first steps don’t feel as meaningless and inconsequential as they otherwise might have.Sign up to the TechRadar Pro newsletter to get all the top news, opinion, features and guidance your business needs to succeed!“When you are getting into programming, the bar needs to be low enough that you can start the process and find enjoyment, so that you can easily learn. And it will take you a long time to learn,” Torvalds said.But he also cautioned about using AI for “serious” work, especially when starting out. The person using the tool needs to know what they want to do, and need to know exactly what the outcome needs to be, if they want to use it properly.Everything other than that is a risk that can come with grave consequences. So, it would be best to start with “toy” projects or personal projects, before using it for anyhitn more “serious”.“I think that you need to be very careful using AI when you're doing something real and important. But I love the concept of AI as a gateway drug into programming,” he concluded.Fighting fire with fireTorvalds also touched on AI-generated vulnerability reports and patches. In mid-May 2026, he described the Linux mailing list as being “almost entirely unmanageable” due to it being filled up with useless AI-generated reports.“The continued flood of AI reports has basically made the security list almost entirely unmanageable, with enormous duplication due to different people finding the same things with the same tools,” he said at the time.“People spend all their time just forwarding things to the right people or saying "that was already fixed a week/month ago" and pointing to the public discussion”.Torvalds now doubled-down on that assessment, saying the problem is that some people simply “don’t care”, and look for flaws in things no one even uses any more.“We are getting a lot more of these random patches. Sometimes for very important and very real security issues, and sometimes for things that nobody has touched in 20 years, because people don't care.”“When you put an AI bot on it, and ask it ‘Hey, is there a memory leak here?’, it will say ‘I don't care about the fact that nobody uses this driver anymore. I'll happily give you all the error reports that I can find’. As a result, the maintainers end up mixing “interesting” and “not interesting” patches.“But on the whole, I think it is improving our code base,” he concluded.The cost of that improvement is maintainer stress.“It is also stressing maintainers out to the point that it's a problem,” he stressed (pun intended). “Yesterday at the maintainer summit, I'd say three quarters of the discussions were about how to make the AI tools - both code generation and the review - less stressful and more successful and more useful for everybody.”“This is clearly a big issue.”The solution, ironically, is not “less AI”, but quite the opposite - more AI.“We are, to some degree, fighting the AI deluge of patches with our own AI deluge of patch reviews,” Torvalds said. “People are jokingly saying that there are bots talking to bots, and it does actually happen.” Follow TechRadar on Google News and add us as a preferred source to get our expert news, reviews, and opinion in your feeds. Sead is a seasoned freelance journalist based in Sarajevo, Bosnia and Herzegovina. He writes about IT (cloud, IoT, 5G, VPN) and cybersecurity (ransomware, data breaches, laws and regulations). In his career, spanning more than a decade, he’s written for numerous media outlets, including Al Jazeera Balkans. He’s also held several modules on content writing for Represent Communications.