The Formal Removal of 'Assisted-by:' and Kernel Documentation Revisions
The Linux kernel development tree has officially eliminated the requirement for contributors to disclose generative AI usage through the 'Assisted-by:' Git commit tag. The tag, which was originally introduced to monitor and document synthetic or assisted code contributions across the kernel codebase, has been phased out across all relevant guidelines.
The policy shift originated with a patch submitted by maintainer Sasha Levin on October 9, 2026. Following substantial debate between October 13 and October 15, 2026, kernel documentation maintainer Jonathan Corbet merged the patch into the docs-next tree on October 14, 2026. The modification removes all references to the mandatory tag from Documentation/process/coding-assistants.rst, submitting-patches.rst, and related localized translations.
Crucially, the tooling enforcement logic was also stripped from scripts/checkpatch.pl. This removes automated warnings during patch preparation, reflecting an engineering consensus that administrative tags cannot practically guarantee provenance or prevent undisclosed machine-generated contributions.
Torvalds Intervenes: Code Comprehension Over Tool Disclosure
The primary catalyst for the repeal was practical evidence presented by maintainer Sasha Levin, showing that the attribution requirement was fundamentally unenforceable. Repository tracking indicated that only roughly 10% of patches utilized the Assisted-by tag, even as the adoption of automated coding assistants grew exponentially among everyday contributors.
Linus Torvalds intervened decisively on the Linux Kernel Mailing List (LKML) on October 13, 2026, rejecting blanket bans and mandatory disclosure bureaucracy. Torvalds articulated that the project's expectation is for contributors to perform at their best and understand their submissions, asserting that genuine engineering comprehension has very little to do with the specific tools employed to produce the patch.
Demonstrating this principle firsthand, Torvalds had submitted his own patch series titled 'noref-filemap-read-with-llm' on October 12, 2026. The series was authored utilizing Anthropic's Claude Code coding agent, accompanied by rigorous manual testing, proving that verified code quality—not tool isolation—remains the sole operational standard for kernel inclusion.
Practitioner Debate: Verification Pragmatism Versus Pull Request Volume
The decision has triggered significant debate across the global software engineering community. Systems programmers and seasoned maintainers largely welcomed Torvalds' stance, agreeing that peer review must evaluate concrete operational safety, architectural fit, and automated test passes rather than relying on the honor system for provenance labeling.
Conversely, conservative maintainers and open-source stewards voiced strong reservations, expressing concern that removing attribution barriers might exacerbate the influx of superficial, hallucinated, or low-quality automated pull requests. For smaller projects lacking the kernel's extensive continuous integration infrastructure, triaging automated contributions presents a severe operational bottleneck.
The Linux kernel's permissive, results-focused approach contrasts sharply with other prominent ecosystems. Projects such as NetBSD, QGIS (under proposal QEP 428), and System76 have established strict prohibitions or heavy disclosure barriers against AI-generated code. Meanwhile, supplementary kernel governance proposals, including draft automation guidelines under AGENTS.md, remain under active review by maintainers.
Strategic Implications for Thai Enterprises and Engineering Leaders
For Chief Information Officers, Chief Technology Officers, and enterprise software teams in Thailand, the Linux kernel's policy reversal offers a critical template for internal AI governance. Imposing bureaucratic disclosure requirements or blanket tool bans often generates friction without providing real auditability, driving tooling usage into unmonitored shadow IT channels.
Thai technology organizations must pivot their compliance strategies from tool policing to rigorous output verification. This entails reinforcing automated CI/CD pipelines, static analysis, unit testing, and architectural review standards. Regardless of whether an internal engineer uses Claude, GitHub Copilot, or local open-weight agents, full professional liability for security flaws or runtime bugs must remain anchored to the human committer.
As regulated sectors in Thailand—such as commercial banking, financial services, and telecommunications—modernize their core infrastructure, technical management should adopt this pragmatism. Ensuring code safety through automated integration harnesses and peer review rigor provides substantially higher institutional resilience than relying on self-reported developer tags.
The governance shift by the foundational open-source project sets a major industry precedent: enterprise software teams should audit code correctness and human accountability rather than relying on unenforceable disclosures.