QGIS AI Policy Takes a Strict Turn
The QGIS AI policy has been formalized through QEP428, defining AI contributions as strictly not allowed. The proposal was merged on October 7, 2026, after maintainers argued that generative AI can increase submission volume beyond what human review can realistically handle, creating either larger backlogs or lower-quality review.
The decision is intended to protect more than code quality; it also protects the contributor relationships on which QGIS depends. Maintainers describe free-software communities as human networks, and organizations value QGIS partly because they can work with real contributors for support, training, and feature development. If maintainers are overwhelmed by submissions they cannot fully assess, the project’s ability to welcome human contributors may suffer.
The QGIS AI Policy Leaves Boundaries in Question
The discussion leaves important questions about what counts as an AI contribution. Participants asked whether the rule covers issue reports, comments, pull requests, and the use of language models for review, auditing, or security work. Some argued that human-verified assistance can help developers understand complex code and find defects. Others warned that AI may give inexperienced contributors false confidence, encouraging patches they do not genuinely understand.
A Pause While the Project Reassesses AI
Maintainers describe QEP428 as a stopgap while the project evaluates AI’s longer-term consequences, rather than as a mature system of exceptions. Possible uses for review, security, or other internal work may require separate discussion. For QGIS, the central challenge is not simply deciding whether AI is good or bad, but creating an enforceable boundary among maintainer capacity, contributor learning, verifiability, and community trust.