Linus Torvalds and Dirk Hohndel held their 30th fireside chat during a keynote session on May 20 at the 2026 Open Source Summit North America in Minneapolis, and this one was packed with substantive technical discourse. The conversation touched on everything from Torvalds's recent Linux 7.1-rc4 release to his hobby projects—including a guitar effects pedal available on GitHub under GPLv2—to the complicated relationship between AI tooling and kernel development. Jim Zemlin introduced Torvalds as the creator of three industry-shaping tools: the Linux kernel, the Git version-control system, and the Subsurface dive-log application. Hohndel quickly noted it might need to be four, given that guitar pedal project.

The AI Coding Boom

Torvalds revealed that something changed roughly six months ago in kernel development velocity. "We've seen a lot more commits, and about 20% more commits for the past two kernel releases than the project had seen for many years," he said. His initial theory was that companies were pushing to get code into version 7.0 because it's a .0 release—similar to what happened with 6.0. "And it turns out I was wrong." The actual driver? AI tooling had improved enough that the barrier to writing a kernel patch has dropped significantly, bringing in more contributors across pretty much all fronts.

Security Report Overload

The flood of AI-generated bug reports has made the kernel's security list "almost entirely unmanageable," Torvalds said. Maintainer Willy Tarreau updated the security bugs documentation with a new policy: if you find a vulnerability using AI, consider it public knowledge—because if one person found it with AI, approximately 100 others likely did too. This represents a major shift in how the kernel community handles responsible disclosure. "We're seeing some of the same effects now with AI," Torvalds explained, drawing parallels to around 2000 when he had to fundamentally change his workflow as Linux scaled beyond what one person could manage. The difference is that this time it's maintainers across thousands of projects who are at risk of burnout when they receive a flood of automated reports from researchers running AI vulnerability scanners.

On AI Tools and Productivity

When asked about tools for code review and patch understanding, Torvalds mentioned Sashiko, which produces reviews of patches sent to the kernel mailing list. He encouraged developers to explore local AI tooling: "You don't want to be entirely at the mercy of the big companies that at some point decide, oh, we need to make money too." But he was quick to distinguish between using AI as a tool versus relying on it entirely for maintenance work. "Even when someone uses AI for coding, if it's for a project that will be maintained for a long time, you need to understand not just your prompts, but you need to understand your end result," he said. His perspective cuts through the hype: AI is changing programming in the same way compilers did—dramatically increasing productivity without fundamentally altering what programming actually is.

The Compiler Comparison

Torvalds offered a characteristically blunt take on AI's revolutionary status: "I claim that compilers increase your productivity by a factor of a thousand. So AI is great, but AI is not changing programming." He noted that people who say 99% of their code was written by AI are missing the point—"I pretty much guarantee that 100% of their code is written by compilers, but they never say that." Even when working on pet projects like his guitar pedal, Torvalds still examines the assembly language output because that's what he grew up with. "It leaves an imprint," he said about writing machine code directly—the numbers, not even assembly. He compared AI to compilers in terms of workflow transformation: "You will all use AI to generate the code that the compilers use to generate the code that the assemblers then use to generate the machine code."

The Guitar Pedal and Hobby Projects

Away from kernel politics, Hohndel revealed he's an early beta tester of Torvalds's guitar-effects pedal project. "It's not bad, and of course there's a 3D-printed housing for it and everything," Hohndel said. "It's really fun." Torvalds deadpanned that he would "change the world of music, too" before warning attendees that they'd need to manufacture the device themselves—either by hand or preferably by sending design files to a PCB manufacturer. The project includes all software and schematics under GPLv2. When Hohndel expressed wonder at finding bugs in applications he uses without understanding how to fix them, Torvalds agreed: "We're actually past the point where people think that open source is just for engineers."

Key Takeaways

  • AI tools have driven roughly 20% more commits to recent kernel releases compared to historical averages
  • New policy treats all AI-discovered vulnerabilities as public because others likely found them too
  • Security researchers running automated scanners often don't stick around to help with fixes—creating maintainer burnout risk across thousands of smaller projects
  • Torvalds's productivity analogy: compilers = 1000x, AI = 10x; both are transformative tools that don't change programming fundamentals
  • Local AI tooling is preferable to depending on large companies' cloud offerings for sustained development work

The Bottom Line

Torvalds gets it. While the industry scrambles to position AI as a paradigm shift in software development, here's the kernel maintainer who's been watching toolchains evolve since writing machine code by hand—telling us that yes, AI is useful, but it's just another layer in the compilation pipeline, not some fundamental reinvention of coding. The real pain points are social: automated vulnerability reports from researchers chasing attention, maintainers drowning in low-quality AI-generated submissions, and the uncomfortable reality that open-source projects without the Linux kernel's resources face existential burnout risk as AI scanning tools proliferate.