Linux 7.4-rc1 arrived on November 1, 2026, with approximately 19,671 commits — the largest merge window in the kernel's history. Buried in that flood is a number that matters more than the total: roughly 3,008 commits were AI-assisted, about 2.4 times the previous merge window's count. The kernel, the most scrutinized and conservatively governed open-source project on earth, just absorbed thousands of AI-assisted contributions in a single release cycle. For founders building on open source, that is not a curiosity. It is a structural shift.
What actually happened
The headline numbers tell the story plainly. Nearly twenty thousand commits in one merge window is an extraordinary volume of work, and the AI-assisted share — around 3,008 commits, up 2.4x from the prior window — shows that AI tooling is no longer a marginal experiment in kernel development. It is a mainstream contributor to the world's most important open-source project.
Two governance details matter as much as the numbers. First, the "Assisted-by" tag marking AI-assisted work was abolished mid-window — contributors no longer must flag AI assistance. Second, Linus Torvalds prefers AI-assisted work in the merge window rather than as release-candidate fixes, and disclosed using Claude himself to prototype a file-read optimization, uncovering three bugs.
Taken together, the community has moved past whether AI assistance is legitimate to how to manage it well. Abolishing the tag is not secrecy; it is a judgment the label outlived its usefulness. What matters is code quality, not keystroke provenance.
Why the kernel is the right test case
Skeptics argue AI output suits boilerplate but not serious systems work. The kernel is the hardest rebuttal: it runs on billions of devices, its maintainers are famously exacting, and code landing there has survived review by elite systems programmers. If AI-assisted commits clear that bar by the thousands, the inherently-low-quality argument is finished.
Torvalds's own Claude use sharpens the point: the kernel's founder did not merely tolerate AI tools; he used one to prototype an optimization and found real bugs. Judge the tool by what it produces, not by ideology about how code should be written.
His merge-window preference is equally practical: merge windows are for new work, release candidates for stabilization. AI-assisted contributions, like any, are welcome when they respect the project's rhythm. The rule is process discipline, not tool ideology.
What this means for open-source founders
For founders building companies on open source — whether maintaining a project, commercializing one, or depending on one — the Linux 7.4 cycle offers several lessons.
Contribution velocity is now a competitive variable. If the kernel can absorb 19,671 commits in a window with a large AI-assisted share, smaller projects can move dramatically faster too. Founders who draft drivers, tests, docs, and refactors with AI help, then review rigorously, will ship faster than competitors writing everything by hand.
Review capacity becomes the bottleneck. AI accelerates writing far more than reviewing, and every AI-assisted commit still needs a human maintainer to read, test, and approve it. Projects scaling AI writing without scaling review will drown in pull requests — invest in review tooling, automated testing, and maintainer capacity.
Labels matter less than outcomes. Communities spent years debating how to label AI contributions; the kernel's answer is that the label matters less than the work — does the code pass review, pass tests, survive in production? Require disclosure where it genuinely helps review, but do not build a provenance bureaucracy that distracts from quality.
Expect the LTS question to matter more. Linux 7.4 is expected to become the 2026 long-term support (LTS) kernel, meaning its code will be maintained and deployed for years. AI-assisted commits landing in an LTS release will be scrutinized over a long horizon, and any systematic quality problems will surface. That is a healthy discipline: the long tail of maintenance is where AI-assisted code will truly prove itself, and founders depending on LTS kernels should watch the stable-tree experience closely.
The bottom line
Linux 7.4's merge window is a milestone, not because of the record commit count but because of what it normalizes. AI-assisted contribution is now ordinary infrastructure work, judged by what lands rather than by how it was written. The kernel did not lower its standards to accommodate AI; it absorbed AI output at scale while keeping its review process intact. That is the model for every open-source project and every founder building on one: use the tools, keep the bar, and let the merged code speak for itself. The projects that learn that discipline will move faster than ever. The ones that treat AI assistance as either magic or contamination will be left behind.
Angeled.com connects founders with investors.
Kernel merge-window statistics on AI-assisted commits (~3,008; ~2.4x prior window; Assisted-by tag abolished mid-window)
Linus Torvalds statements on AI-assisted work in the merge window and his use of Claude to prototype a file-read optimization (3 bugs uncovered)
Expected 2026 LTS kernel designation for Linux 7.4
Read Next
OpenAI Releases Codex Security as Open-Source CLI and SDK
Databricks Unveils Always-On Genie, an AI Agent That Stops Waiting to Be Asked
DeepMind Launches AlphaGenome 2, a DNA Model That Reads Genomes Cell by Cell
ElevenLabs Unveils 13 New Products and Upgrades at Its New York Summit
Google Launches Gemini Decision 1, a Model That Answers Instead of Writing
Runway Open-Sources Praxis-1, Its Robot Model That Learned by Watching Video



