The creatures born from code: Generative AI threatens to upend citizen science records
You need to tread carefully among the leaf litter of the Indonesian scrubland and its tropical rainforests. The dead leaves might seem caught in a gentle breeze, but there's more here than meets the eye. Try plucking one from the ground and you'll find an insect caught in the act of prayer.
In the leaf litter of Indonesian scrubland and tropical rainforests, the dead leaves conceal not just life, but a sinister threat. Kris Anderson, a mantis specialist, discovered a Deroplatys, a type of dead leaf mantis, in a citizen science platform called iNaturalist. However, upon closer inspection, Anderson realized the specimen was a fabrication, created using generative AI image tools.
As these AI tools become more sophisticated, the risk of fake biodiversity records increases, threatening the integrity of citizen science initiatives like iNaturalist and eBird. Anderson has documented this vulnerability, warning that fake observations could infiltrate scientific papers and assessments.
The problem is already apparent. In November 2025, a willow tit in Scotland appeared on social media as if feeding from a bird feeder, despite being an extremely rare species. Upon closer examination, it was revealed to be a fake. This incident highlights the potential for AI to contaminate scientific records.
Researchers at Manchester Metropolitan University, including bird specialist Alexander Lees, have warned that AI-generated data could taint biodiversity databases. While creating entirely AI-generated observations is problematic, enhancing existing images is also a concern. The team demonstrated how easily a fake bird could be generated, with amusing results.
However, the dangers are more serious. A red-winged blackbird image, created by asking AI to improve an existing image, was accepted as a legitimate sighting in Brazil. Later, the image was identified as a fake and removed from the database. This incident underscores the potential for AI to manipulate scientific records without detection.
History has shown that such deceit is not new. The Charlton brimstone, described by James Petiver in the early 1700s, was actually created by hand to mimic wing markings. This highlights the need for vigilance in verifying biodiversity records, as AI-generated fakes could become commonplace before adequate safeguards are in place.
Written by urgent.news from Phys.org's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
