Amid Surge in AI-Generated PRs, Open Source Maintenance Crisis Centers on 'Ownership' and 'Context'

Over the past year, the surge of AI-generated PRs in open source projects has been increasing the burden on maintainers. Unlike spam of the past, today's AI-generated code passes tests and appears formally correct, but it lacks ownership and context, which raises the actual cost of maintenance.

Lack of Ownership and the Context Gap

AI has lowered the cost of producing code, but it has not lowered the time and skill cost required to understand a project. Unlike in the past, when contributors learned project norms through PR feedback, AI does not go through social interaction, so the context gap does not narrow. This leads to problems such as the reintroduction of trade-offs that had already been rejected and an increased future maintenance burden, with maintainers reporting cognitive load.

Shift Toward Machine-Readable Infrastructure

The open source ecosystem has followed a pattern of evolving from social norms to explicit guidelines to enforced checks. AI-assisted contribution is predicted to follow this same path. This requires building infrastructure that increases machine readability, such as specifying architectural constraints in .md files within repositories. This is not mere documentation but infrastructure for AI agents, enabling pre-implementation validation.

Strengthening Accountability and Reaffirming the Social Contract

Using AI does not reduce accountability—it increases it. Every PR should have a clearly accountable human owner, who is responsible for understanding the code, explaining its necessity, responding to review feedback, and fixing errors. An enforced code of conduct that covers contribution quality and accountability is now essential, and it must be reaffirmed that contribution—even when using AI—remains a social relationship built on trust.