Linux Kernel Sets Groundbreaking AI Coding Standards: What Developers Must Know

On 12 April 2026 the Linux kernel community added an official AI‑assisted development policy in the file Documentation/process/coding-assistants.rst. The document defines how AI‑generated contributions must be identified and who bears responsibility.

Two core rules

AI cannot use the Signed‑off‑by tag

The Signed‑off‑by line is a legally binding developer declaration. Only a human may add this tag; AI agents are prohibited from signing, ensuring AI contributions cannot be presented as human authorship.

Mandatory AI attribution

All AI‑assisted patches must include a new “Assisted‑by” tag with a fixed format:

Assisted-by: AGENT_NAME:MODEL_VERSION [TOOL1] [TOOL2]

AGENT_NAME is the AI tool name, MODEL_VERSION the specific model version. Optional analysis tools (e.g., coccinelle, sparse) may be listed; basic development tools (git, gcc, make) need not be mentioned.

Example from the kernel documentation:

Assisted-by: Claude:claude-3-opus coccinelle sparse

Claude is cited because it is already widely used in the kernel community for code review and rule generation, making it the first AI model referenced in official Linux documentation.

Responsibility chain

The human submitter remains fully responsible for any AI‑generated code, must verify GPL‑2.0‑only compliance, and sign the patch with their own Signed‑off‑by. The AI model or its vendor bears no liability.

Industry implications

The guideline provides a concrete, traceable, and enforceable framework for AI‑assisted development, balancing efficiency gains with legal and security safeguards. It is expected to become a reference model for other open‑source projects facing similar AI integration challenges.