QGIS Implements Total Ban on AI-Assisted Contributions

QGIS Adopts Strict No-AI Policy

The QGIS project has formally amended its development policy to prohibit all AI, LLM, and agent-assisted code contributions to the qgis/qgis repository. This change, enacted via QEP428 (QGIS Enhancement Proposal), replaces the project's previous "human-in-the-loop" policy, which had required transparency and mandatory human review for AI-generated code. The new policy mandates that any contributions found to be created via AI or agent-based workflows are subject to immediate code reversion and potential sanctions against the contributor.

Rationale for the Ban

Maintainers cited three primary reasons for the shift to a total ban:

• Unbearable Maintainer Pressure: The volume of AI-assisted pull requests has created a significant burden on the project's volunteer maintainers, who were already managing a high workload before the rise of LLMs.
• Quality and Review Sustainability: The influx of AI-generated code often lacks the depth of understanding required for a complex project like QGIS. This leads to either a backlog of unreviewed code or the integration of code that has not been thoroughly vetted, both of which threaten the project's stability.
• Community Integrity: The project relies on a human network of contributors who provide support, training, and features. Maintainers expressed concern that AI-assisted contributions bypass the learning process, preventing new developers from building the necessary skills to eventually become reviewers themselves.

Future Outlook and Implementation

Project leadership has characterized this policy as a "stop-gap" measure rather than a permanent, immutable rule. The goal is to start with a "closed door" to protect the project's current stability, with the possibility of selectively reopening it for specific use cases—such as IDE autocomplete or language translation—on a case-by-case basis in the future.

While some community members have expressed concern regarding the policy's broad wording and potential for enforcement challenges, maintainers emphasized that the measure of compliance relies on the contributor's conscience. As one maintainer noted, "You will know if your brain made the code you want to share or someone else's brain did."

Community Perspectives

Discussion surrounding the policy has been extensive, reflecting a broader tension in open-source development regarding the role of automated tools:

"I’m clearly not yet skilled enough to fix bugs in a project as complex as QGIS, but LLMs give you the illusion that it’s possible. A few tokens later, and the bug is fixed. By doing this, we’ve broken a trust contract between the reviewer and the contributor." — Amalvir, contributor

"I am afraid that they might throw out the baby with the bathwater... I have found LLMs to be useful for discovering additional, non-obvious context in the code. They are often wrong, but it is my strong belief that the results can be used by humans to gain further understanding." — kannes, contributor

"The bottleneck is no longer writing code but staying aligned with it, knowing what it does and whether it does what you meant. In my own work, that’s now most of the job, even for code I asked for. For maintainers reviewing large PRs from strangers, it's certainly much worse." — jcolot, Hacker News commenter