Developer open-sources AI-generated bird stickers, sparking misinformation concerns

On November 11, 2026, ifanr reported on a project called the "Hua Shu Global Bird Sticker Library", which set off a dispute over AI-generated science communication. According to the report, an independent developer who says he cannot code at all built a set of bird stickers as supporting material for his own birdwatching app. Working with AI scripts and so-called Vibe Coding, he burned through more than 1 billion tokens and exhausted the weekly quota of a $200 top-tier membership several times over, generating outline-style stickers of the world's birds within about a week and announcing the set as officially open source. The project initially covered 11,064 living bird species, later adjusted to 11,011 species and 91 subspecies after the checklist was updated, and it also launched a multilingual bird website supporting 15 languages.

The controversy centers on two things: where the reference images came from and how accurate the results were. The report says the developer wrote scripts to scrape photographs from iNaturalist and Wikimedia Commons in bulk to use as AI reference images, even though many of those photos are unlicensed or explicitly disallow derivative works. In the author's own testing, quite a few stickers showed anatomical errors such as wrong iris colors and fused toes, yet 3388 images that fell below his own stated quality threshold were still included in the release.

Birdwatching communities criticized the approach of feeding images to the AI first and taking them down only after complaints, while the article compared the situation to a passage from Lu Xun's essay 藤野先生, in which an anatomy diagram must not be altered to look prettier. It concluded that rigorous natural science should not be treated as an AI skills competition. At this point the episode remains at the stage of community criticism and the author's own public retrospective; the report does not say whether the project has changed course since.