Godot's maintainers just drew a hard line against the flood of AI-generated code slop drowning their volunteer review process. The Foundation's June 30 policy update bans autonomous agents, vibe coding, and any substantial AI-written code in pull requests, citing the demoralizing waste of time spent dissecting low-quality submissions that contributors often don't understand themselves. Only menial tools like basic autocomplete or regex get a pass, with full disclosure required if AI touches anything more.

The move follows months of complaints, including from veteran Rémi Verschelde who called the AI PRs "draining and demoralizing" back in February as the volume exploded and reviewer bandwidth stayed flat. The core issue isn't just quantity—it's that feedback loops break when code isn't human-authored, killing the mentorship pipeline that turns new contributors into future maintainers. All PRs still need human review and approval before merging, and AI-generated text in communications is out too.

This isn't a full rejection of AI assistance, but a clear signal that open-source projects like Godot won't outsource accountability or quality control to models that can't fix their own messes. Studios pushing genAI elsewhere in the industry might want to take notes before their own contributor bases revolt.