Maintainers and readers push back on AI-made work

QGIS banned AI-assisted contributions outright, and the thread immediately got stuck on the definition problem. One commenter still writes code by hand but prototypes with AI first, and asked whether that counts. Another dismissed the maintainers as "threatened by AI" and suggested cloning QGIS from scratch with agents. In the Doory thread (an open-source Canva), a commenter was "horrified" to find the whole project authored by Claude and spawned from a single commit. A separate commenter noted that Canva's real moat is its templates, assets and stock photography, not the editor.

The same fatigue shows up elsewhere. Someone skipped the Postgres storage article because the title read as LLM-generated. The Ask HN about adding a flag for AI-generated articles ran into three problems: detectors are unreliable, voting can be gamed, and HN's new guideline ("Don't post generated text or AI-edited text") only clearly covers comments.

The pattern here: the cost of producing a plausible repo, article or PR is now near zero, so the scarce thing is provenance. Communities are reacting by adding gates, and the gates are crude because nobody can define "AI-assisted" cleanly.

The counterpoint is that bans push away people who use AI responsibly, and one commenter said a similar mood made them feel unwelcome on Lobsters.