Linux creator Linus Torvalds tells AI haters to walk away from Linux, or go fork it
The New Stack Adrian Bridgwater ● Covered by 2 sources
Linus Torvalds says Linux is officially pro-AI now, telling critics to fork the project if they don't like it. Just 20 months ago he was calling AI 90% hype—so this is a real about-face from the top of the kernel.
Based on reporting by The New Stack, Adrian Bridgwater — read the original for the full story.
Summary, retelling and take written by AI under human oversight; images are AI-generated illustrations. How we work · Report an error
Linus Torvalds doesn't do wishy-washy, and his latest post on the kernel mailing list proves it. As the top-level maintainer of Linux, he's decided the debate over AI-assisted code is closed. Not up for discussion, not up for a vote among grumbling contributors. If you hate that AI tools are now welcome in the codebase, he says, you can fork the project or walk away. Those are the options.
What makes this notable isn't just the bluntness — it's the reversal. Back in October 2024, Torvalds told TFIR he was basically ignoring AI because the whole industry felt like "90% marketing and 10% reality." He figured it would take five years before anyone could tell what AI was actually good for versus what was just a flashy ChatGPT demo. It's been 637 days since he said that. Apparently he didn't need the full five years.
His current position is more nuanced than a simple thumbs-up, though. He calls AI a tool, useful like any other tool engineers rely on, but also admits it's "a somewhat painful" one — partly because of the extra burden it dumps on maintainers, and partly because it keeps surfacing bugs that are frankly embarrassing. His answer isn't to ban the tools or pretend they don't exist. It's to make the tools serve maintainers instead of drowning them in work, which is a much harder problem than just saying yes or no to AI.
Others in the industry are backing him up. Luke Hinds of nolabs argues the code-quality argument was always a red herring — what actually matters is whether a human submitting a patch understands it well enough to explain the intent behind it. He's already seeing the failure mode play out: pull requests padded with thousands of lines of AI-generated justification for changes nobody actually reasoned through. Red Hat's Mike McGrath strikes a similar note, framing AI as a collaborator for engineering problems, with human oversight as the non-negotiable guardrail.
So Torvalds hasn't gone soft on standards. He's just decided AI is a fact of the workflow now, the same way Git or compilers are facts of the workflow. What he still cares about, loudly, is whether the person hitting submit actually understands what they're submitting and why it belongs in the project. That part hasn't changed one bit — only his patience for people arguing against tools that are already here.
My take — AI-written commentary, not fact-checked reporting
Torvalds flip-flopping this fast just confirms what a lot of us already suspected: the 2024 AI skepticism from serious engineers was never really about the tech, it was about the marketing noise drowning out the tech. Once the noise settled and the tools started actually shipping usable code, the holdouts folded — and Torvalds folding publicly, with his usual lack of diplomacy, is basically the industry's permission slip. The people still fighting this battle at the maintainer level aren't defending code quality anymore, they're defending a vibe, and vibes don't survive contact with a top-level maintainer who's out of patience.
Read more about this at: The New Stack
