"I think the whole 'don't use AI for commit messages' kind of argument … Maybe it's well-intentioned, but I think it's entirely misguided," Linus Torvalds told the author in a recent exchange, a succinct restatement of a position that has set off sharp reactions in some corners of the open-source world.
Linus Torvalds: competent people may — and should — use AI
Torvalds has moved from caution to practical acceptance: he says AI is a useful tool in the hands of developers who understand the code they are changing. The kernel maintainer drew a clear distinction between skilled contributors who use AI to assist their work and "clueless people using AI" who submit patches without understanding the code. Torvalds called for discouraging "drive-by patches" while welcoming contributors who use AI as part of a learning path toward becoming "useful members of the kernel community."
Deprecation of the Assisted-by tag at the Linux Kernel Maintainers Summit in Prague
At the Linux Kernel Maintainers Summit in Prague, Sasha Levin announced that maintainers had decided to deprecate the Assisted-by tag. Under that change, developers no longer need to declare whether they used AI in their submissions. The deprecation does not, however, change quality expectations: Torvalds emphasized that AI-assisted code will still be judged by the same standards—sloppy or incorrect contributions will not pass muster.
Audit-ready is a season. It shouldn't be.
Evidence in spreadsheets, controls drifting between audits, frameworks multiplying on flat headcount. Nubivance runs continuous compliance on Rapid7 Cyber GRC - SOC 2, HIPAA, ISO 27001, PCI, CMMC.
End the scrambleTwo distinct AI problems for Linux maintainers
Torvalds and other senior maintainers framed the AI-related challenges as two separate issues. First, automated tools are finding real bugs in a very large codebase, producing more reports and more work for already "fairly overworked" maintainers. Second, there are human-behavior problems: people using generative tools to create patches without the requisite understanding. Torvalds said both are "very much both very real problems," but argued they are not problems with AI itself so much as problems exposed by AI.
From "AI slop" to useful bug reports — Greg Kroah-Hartman's experience
Long-term Linux kernel maintainer Greg Kroah-Hartman described the shift in the quality of AI-produced outputs. Months ago, he said, maintainers were getting "AI slop"—security reports that were "obviously wrong or low quality." Then, "something happened a month ago, and the world switched. Now we have real reports." Kroah-Hartman and others acknowledged that the influx of useful bug reports is a double-edged sword: the reports are valuable, but they also risk burying maintainers under a new volume of issues to triage.
AI tooling and the hardware-driver problem: hw-harness and reverse engineering
Not all AI effects are burdens. Christian F.K. Schaller, Red Hat's director of software engineering, created hw-harness while working on a Linux driver for a slide scanner. The open-source project provides tooling to determine how a piece of hardware behaves so that a developer can write a Linux driver. By automating slow, tedious methods used to reverse-engineer binary-only Windows drivers, hw-harness makes building Linux drivers from Windows drivers easier—potentially addressing long-standing gaps in desktop hardware support.
What this means for kernel maintainers, new contributors, and hardware developers
- Kernel maintainers: Expect a sustained increase in the volume of bug reports and patches; useful AI-generated reports can surface real issues but also add to triage workload, as Kroah-Hartman and Torvalds both warned.
- New contributors: AI can act as a "gateway drug" to programming, Torvalds said at Open Source Summit Europe, enabling newcomers to do things they would otherwise find very hard—provided they move beyond "vibe-coding" to learn the codebase.
- Hardware developers and driver maintainers: Tools like hw-harness can speed reverse-engineering of proprietary drivers and ease the creation of open drivers, potentially improving device support on Linux desktops.
Public reaction has been mixed. Critics ranged from Roy Schestowitz of Tux Machines, who wrote "Linus Torvalds Turns to Slop," to podcaster David Gerard calling it "Linux kernel goes full AI slop code," and anonymous commenters on Tom's Hardware alleging influence from the Linux Foundation. Torvalds and other senior maintainers reject the notion that adoption equals abandonment of quality. Their message is pragmatic: use AI where it helps, police submissions that don't demonstrate understanding, and expect the maintainers' workload to change as tooling evolves.
Torvalds summed the stance succinctly: AI is fine for competent developers, but the community must "discourage the drive-by patches" that amount to little more than illusionary contributions. The near-term question, grounded in the facts these maintainers themselves report, is whether the community can absorb both the benefits and the surge of work that better AI tooling is bringing to an already overworked corps of maintainers.
