Software · 17 July 2026 · 3 min read
“Fork it or walk away”: Linus Torvalds shuts down anti-AI crusade in Linux development
In brief: Debates over Sashiko, an autonomous AI code review agent for the Linux kernel, sparked protests from maintainers opposed to automation. Linus Torvalds shut down the critics, telling those who reject AI tools to fork the project or walk away, and defining artificial intelligence as a useful tool based on technical merit.
by Team Mocchi's
The debate over the integration of artificial intelligence into software development has reached the very heart of open source: the Linux kernel. Faced with efforts by certain factions to outright ban the use of code generated by large language models, project creator Linus Torvalds has taken a sharp, uncompromising stance. With his trademark pragmatic and direct style, Torvalds shut down the anti-AI crusade, declaring that Linux will not shut its doors to technological innovation for ideological reasons.
The Sashiko controversy and maintainer overload
The discussion erupted within the official kernel developers' mailing list following trials of Sashiko. This agentic code review system is designed to analyze code changes and detect bugs before they ever reach production. According to data provided by its creators, Sashiko can independently identify over 53% of the issues that a human developer would only fix later in the development cycle.
However, as reported by Ars Technica, the introduction of this agent met with significant backlash. The friction lies in the tool’s false positive rate, which sits around 20%. Many kernel maintainers complained about being flooded with automated, irrelevant bug reports that eat up valuable time otherwise spent on manual code review. This tension quickly transformed a purely technical issue into a battle of principles regarding the legitimacy of AI in programming.
The ideological clash and the SFC's appeal
The protests of some contributors found support in more organized quarters of the open-source community. During the debate, a participant cited a recent statement from the Software Freedom Conservancy (SFC). The organization expressed deep skepticism toward LLMs, arguing that the free software community should support, rather than just tolerate, those who outright reject generative AI systems, ensuring every contributor maintains self-determination over their code.
This perspective, which seeks to erect ethical and operational barriers against algorithms, directly collides with the reality of productivity tools that many developers now find indispensable for accelerating routine tasks.
Torvalds' pragmatism: "AI is just a tool"
Torvalds' response was swift and definitive. In a statement destined to echo throughout the software industry, he made it clear that the Linux kernel will not adopt dogmatic stances: "Linux is not one of those anti-AI projects, and if somebody has issues with that, they can do the open-source thing and fork it. Or just walk away."
Torvalds reiterated that his position is based entirely on technical merit and pragmatism, rather than fear of change. "AI is a tool, just like other tools we use," wrote the Linux creator, adding that its usefulness is no longer up for debate and that anyone who doubts it has likely never actually used it. While acknowledging that machine-generated output is far from perfect and requires oversight, he stated he will "very loudly ignore" anyone attempting to block other developers from utilizing these technologies to enhance their workflows.
Mocchi's take
The firm stance taken by Linus Torvalds offers a valuable lesson for businesses navigating digital transformation and custom software development. Integrating intelligent agents into development pipelines is no longer an ideological choice, but an operational necessity for efficiency and code quality. However, the Sashiko controversy proves that AI deployment cannot be reckless: a 20% false positive rate risks overwhelming teams, wiping out the productivity gains of automation. For businesses, the key challenge is not to reject technology, but to govern it through tailored integration that effectively balances human oversight against algorithmic noise.