The Kyverno project introduced an AI Usage Policy in response to a surge of AI-generated pull requests that overwhelmed maintainers. The core argument is that while AI scales code generation infinitely, human review capacity does not. The post distinguishes between AI as acceleration (amplifying understanding) versus AI as substitution (bypassing understanding), arguing the latter creates technical debt and deprives contributors of growth. Key principles include requiring contributors to 'own their commits' even when AI-assisted, mandating disclosure of AI usage for licensing and trust reasons, and exploring AGENTS.md-style configurations to guide AI tool behavior in repositories. The post frames responsible AI integration in open source as a shared governance challenge requiring community-driven guardrails rather than outright rejection or uncritical acceptance.
Table of contents
Where this conversation beganThe maintainer reality few people seeAI boomers, AI rizz, and the reality of changeAI as acceleration vs. AI as substitutionOwnership still matters — perhaps more than everDisclosure as trust infrastructureWhen AI meets open source: Kyverno’s approachAI-friendly does not mean AI-unboundedDiscomfort, growth, and privilegeA shared challenge across open sourceAn invitation to the ecosystemAcknowledgements and influencesSort: