Linus Torvalds has released Linux 7.1-rc5, the fifth release candidate in the Linux 7.1 development cycle, on May 24, 2026. While the release continues the kernel’s steady march toward a stable mid-June launch, Torvalds used the accompanying announcement on the Linux Kernel Mailing List (LKML) to voice clear frustration with what he sees as an oversized and poorly timed flood of minor patches — many of them produced with the help of AI coding agents.

AI Coding Agents Are Reshaping the Kernel Patch Flow

Linux 7.1-rc5 continues a pattern first confirmed during the rc2 and rc3 stages: AI coding tools such as GitHub Copilot and Claude Code are now directly contributing fixes that land in the kernel tree. According to Phoronix, these AI-assisted fixes span a wide range of the kernel’s codebase — from graphics driver bugs to security issues in C language code. The networking fixes merge for this week even noted that “the craziness continues with no end in sight.”

This is not entirely new territory. Torvalds acknowledged as far back as rc2 that the unusually high patch volume was “likely due to the popularity of AI development tools,” and by rc3 he confirmed it was no longer a temporary spike but the new baseline — a structural shift in how contributions flow into one of the world’s most critical open-source codebases.

Torvalds: The Size Is “Not Entirely Happy”

In his rc5 announcement, Torvalds was direct about his dissatisfaction. The release candidate is considerably larger than an rc5 traditionally is, and while much of that is composed of small, individually low-risk driver changes, the aggregate effect is a kernel that keeps expanding at a stage of the cycle when it should be narrowing.

“To the surprise of absolutely nobody by now, rc5 is pretty big. Quite a bit bigger than rc5’s have traditionally been. I’m not entirely happy about it — most of this is totally trivial stuff to random drivers, which obviously makes it all less scary, but at the same time I’m really not convinced the churn is worth it at rc5 time.”

— Linus Torvalds, LKML, May 24, 2026

Torvalds went on to argue that many patches labelled as “fixes” are simply too minor to justify merging at the rc5 stage. His preference is for such changes to sit in the linux-next tree and be integrated in the next merge window, rather than adding risk to the current stabilisation cycle. He stated plainly that he intends to be stricter about accepting pull requests that do not address genuine regressions, and that some of the problematic pull requests in this cycle were initiated by AI reviewers.

“These things are ‘fixes’, sure, but at the same time a lot of them are simply so irrelevant that I think they’d be better off in a linux-next tree and get merged during the merge window. They’re not regressions, and some of these PRs were AI-initiated.”

— Linus Torvalds, LKML, May 24, 2026

Earlier in the cycle, during the rc4 announcement on May 17, Torvalds had also flagged a separate but related AI problem: the kernel’s private security mailing list had become “almost entirely unmanageable” because multiple researchers were independently filing identical vulnerability reports — all discovered using the same AI tools running against the same code. He called this “pointless churn” and noted that bugs found via automated tools are, by definition, not secret and should not be routed through the private security channel.

What’s Actually in rc5

The release itself is not without substance. Key technical highlights include:

▸
x86 Platform Drivers: Expanded support for HP and ASUS laptops, continuing the hardware enablement work from earlier release candidates.
▸
Intel P-State & AMD P-State Fixes: Multiple driver fixes addressing power-management and frequency-scaling correctness for both Intel and AMD platforms.
▸
Graphics Driver Fixes: Various GPU driver corrections, including some contributed by AI coding agent-assisted patches.
▸
Sound Subsystem: A large volume of sound subsystem fixes were also merged this week alongside the other driver updates.
▸
Networking & Core: Additional networking corrections and miscellaneous core fixes round out the release.

What Torvalds Is — and Is Not — Saying

It is worth being precise about Torvalds’s position. He is not opposed to AI tools in kernel development. As noted during the rc4 cycle, his view is that AI tools “are great when not causing unnecessary pain and pointless make-believe work.” The concern is specifically about low-value patches being submitted too late in the stabilisation cycle, and about AI-generated bug reports creating redundant work for maintainers — not a blanket rejection of AI-assisted contributions.

⚠ Editorial note

Some reporting on this release has attributed strong categorical statements to Torvalds — such as a declared policy of rejecting all AI-initiated pull requests — that go beyond what the LKML announcement actually states. His position is a call for better judgment from contributors, not a formal ban. The quotes and characterisations in this article are drawn from primary sources.

What’s Next

The Linux 7.1 stable kernel is currently expected to ship in mid-June 2026, either around June 7 if the cycle wraps at rc7, or June 14 if an additional rc8 is needed. Torvalds’s stricter posture toward late-cycle pull requests may help rein in the release candidate size over the coming weeks, though the broader structural question — how the kernel community manages AI-accelerated contribution volume at scale — will continue well beyond this cycle.