A Gemini-Made Crocodile Photo Shut Down a Singapore Reservoir for Two Days — Now Its Creator Faces 10 Years

The photograph was not of a real crocodile. It was generated using Google's Gemini app by Ye Lin, a 30-year-old Myanmar national employed by a contractor that performs reservoir maintenance for PUB ^[1][2]. He created the image and sent it to a colleague, Min Min Oo, according to court documents ^[2].

PUB initially ran the photo through multiple AI-detection tools but got inconclusive results, which is why it erred toward a full precautionary search rather than dismissing the tip ^[1]. The agency only confirmed the image was fake the following day and referred the matter to police ^[3]. Investigators allege Ye Lin then deleted the AI-generated image and his Gemini app history from his phone in an attempt to cover his tracks — the act that produced the obstruction charge ^[1][2].

This wasn't an organized disinformation campaign — it was a single fabricated image sent between two colleagues that leaked into an official safety channel and set off a real municipal response ^[2]. That's precisely what makes it dangerous: no coordinated bad actor is required for generative AI to trigger a costly, resource-draining emergency response, just one person with a phone app and a credulous recipient.

FALSE. There was no crocodile at Pandan Reservoir — the photograph was AI-generated, and water activities resumed on August 22 once PUB established that ^[1][3]. Ye Lin was formally charged on September 29 with communicating a false message (up to 3 years in prison and a S$10,000 fine) and obstructing the course of justice (up to 7 years), for a combined maximum exposure of roughly a decade behind bars ^[1][2]. He has indicated he intends to plead guilty without legal representation; his case returns to court on November 10 ^[1]. Singapore police have explicitly warned the public not to submit AI-generated or digitally altered images to authorities ^[3].