QGIS, the open-source mapping software used by researchers, city planners, and hobbyist cartographers, says no to AI-written code.
A pull request filed to the project's QGIS-Enhancement-Proposals repository on GitHub formally bans AI-assisted contributions to the codebase. The proposal, numbered PR #387, quickly traveled from GitHub to Hacker News, where it racked up 68 points and 60 comments. QGIS runs largely on volunteer maintainers who review every submitted patch by hand. A blanket ban removes the ambiguity of reviewers trying to guess which code was machine-generated versus human-written.
This matters because QGIS isn't a toy project - it's foundational software that public agencies, universities, and researchers rely on for mapping and spatial analysis. For a codebase that size, maintainer bandwidth is finite, and AI tools can generate code far faster than a small volunteer team can vet it. A hard rule is a blunt but simple way to keep the review queue from drowning in low-effort, unvetted submissions.
Expect this fight to keep recurring across open-source projects. Maintainers are stuck between wanting more contributions and needing to trust that someone - or something - actually wrote and tested the code they're being asked to merge.