Linux kernel maintainers publish rules for AI-generated code

Linux kernel maintainers released a guide for accepting code produced by AI agents. The document requires evidence of the bug's cause, local testing before submission, and mandatory human sign-off taking full ownership of changes.

• Agents must read full project documentation, not search by keywords
• Evidence of the bug's cause is required, not model assumptions
• Testing and revisions happen in a local loop before submission
• All sign-off certifications must come from humans, not AI