The New Reality: More Collaboration, Same Number of Maintainers
The maintainer workload crisis in open source is the growing mismatch between exploding contribution volume—much of it aided by AI-generated code—and the relatively flat number of people qualified and willing to review, merge, and steward those changes over time. That imbalance is reshaping project policies, tooling, and norms across platforms. Open source maintainers now sit at the chokepoint of a global coding boom, and the big story is not that AI is pushing newcomers out, but that it is amplifying the flow of code into queues that were already overloaded. Outbound collaboration on public repositories grew 16% in a single quarter from late 2025 to early 2026, one of the fastest jumps since the pandemic spike. Merged pull requests have nearly quadrupled, from about 25 million per month in early 2023 to roughly 90 million per month today. That sounds like a victory for open source, but it lands as more issues, more comments, and more complex pull requests that maintainers must handle without a matching surge in their ranks.
Godot’s Hard Line on AI-Generated Code
The Godot Engine’s new rules are a blunt answer to the mounting cost of AI-generated code review. On June 30, the project banned substantial AI-generated code and tightened contributor requirements, explicitly to reduce a review backlog that had become unmanageable. The Godot Foundation says an influx of AI-generated submissions worsened an already growing queue, driven by strict review standards and a shortage of qualified reviewers even as community interest rose. Instead of welcoming every AI-assisted feature, Godot is trying to slow the stream and raise its signal-to-noise ratio. Contributors with three or fewer merged pull requests must now get approval before submitting new features or major refactors and are steered toward bug fixes and documentation first. This is a controversial stance, but it reflects a clear judgment: in the current climate, protecting maintainer time and code quality matters more than maximizing contributor freedom.
AI Coding Agents: Not Killing Newcomers, But Overloading Queues
A new study from Peking University undercuts the popular fear that AI coding agents are crowding out human newcomers—while confirming that maintainers are bearing the real cost. Researchers tracked 1,888 GitHub repositories that adopted tools such as Cursor and Claude Code and compared them to similar projects that had not. Using a difference-in-differences design, they found that newcomer participation stayed level or rose slightly, with the worst estimate showing only a 1.5% decline, too small to be statistically meaningful. The code itself, however, is changing. Cyclomatic complexity rose 3%–4% across languages after adoption, and Python cognitive complexity jumped in the double digits. More importantly, pull request volume has nearly quadrupled, and the people sending those AI-assisted changes may not fully understand what they are proposing. All of that lands on open source maintainers whose numbers are not growing at the same pace, turning review queues into the new bottleneck.
Platform-Level Controls: GitHub Tries to Stem the Tide
If Godot is tightening rules at the project level, GitHub is responding at the platform level to GitHub contributor workload that has exploded. The company describes today’s environment as an "Eternal September of Open Source," a long wave of newcomers that keeps reshaping community dynamics. Rather than betting on fewer contributions, GitHub is giving maintainers new brakes and filters for their repos. Maintainers can now cap how many open pull requests an outside contributor can have at once and restrict pull request or issue creation to collaborators only, or temporarily turn either off. Noise-reducing banners nudge people toward reactions instead of cluttering threads with "+1" comments, while pinned issue comments and faster diffs and navigation make triage less painful. It is an admission that the default of "infinite contributions to every public repo" no longer works when pull requests are measured in tens of millions per month.
Where Open Source Goes Next: Gatekeeping or Sustainable Stewardship?
The immediate takeaway from this moment is stark: AI coding agents and global collaboration are not killing open source participation—they are testing whether open source maintainers can keep projects reliable under constant load. Godot’s ban on substantial AI-generated code is one answer; GitHub’s new maintainer controls are another. Neither is neutral. Both shift power toward maintainers and away from casual contributors, and that shift will define the next phase of open source culture. The Peking University team argues that future work should examine projects born with AI and the intensity of tool use, because current methods cannot distinguish light experimentation from heavy reliance. In practice, the direction is clear. Projects that want to survive will need explicit contributor gates, clearer review expectations, and a pipeline that turns active contributors into maintainers. The choice is not between openness and control, but between unmanaged overload and sustainable stewardship.
