Summary
- An AI-generated image of a crocodile at Pandan Reservoir triggered a search operation and the suspension of water activities.
- A 30-year-old man faces charges of communicating a false message and obstructing the course of justice.
- The offences carry maximum prison terms of three and seven years respectively.
- PUB later revealed that the AI detection tools it initially used produced inconclusive results.
An AI-generated image of a crocodile was enough to trigger suspended activities, a search operation and ultimately criminal charges in Singapore.
The case began on August 20, 2026, when national water agency PUB was alerted to a possible crocodile sighting at Pandan Reservoir. Members of the public were advised to exercise caution, water activities were suspended, and PUB and the National Parks Board (NParks) launched search and capture operations. The following day, the photograph was found to be fake and the matter was referred to the police.
The image was AI-generated
According to the Singapore Police Force, preliminary investigations found that a 30-year-old man had created an AI-generated image depicting a crocodile at Pandan Reservoir and sent it to a colleague.
Court documents cited by CNA identify the accused as Ye Lin, a Myanmar national. PUB later said he worked for its reservoir maintenance contractor and that his supervisor was the person who alerted the agency to the image.
According to the charge, Ye Lin allegedly sent the AI image to a colleague at about 3pm on August 20. The following day, he allegedly deleted the AI-generated images and the Gemini app log history from his mobile phone, which forms the basis of the obstruction of justice charge.
AI detectors could not provide a clear answer
On October 1, PUB revealed that it had used several AI detection tools to examine the image immediately after the report. The results were described as “inconclusive”, meaning the tools did not provide a clear answer on whether the image had been generated with artificial intelligence.
The agency therefore took precautionary measures in the interest of public safety. PUB and NParks carried out the search operation, while the Singapore Rowing Association and Singapore Canoe Federation were told to suspend all water activities at the reservoir.
The detail is particularly relevant to photographers, journalists and visual creators: a result from an AI detector is not, by itself, proof that an image is authentic or manipulated. File provenance, context and verification against primary sources remain critical.
The two charges
Ye Lin was charged on September 29 with two separate offences.
The first is communicating a false message under Section 14D of the Miscellaneous Offences (Public Order and Nuisance) Act 1906. The offence carries up to three years in prison, a fine of up to S$10,000, or both.
The second is obstructing the course of justice under Section 204A of the Penal Code 1871, which carries up to seven years in prison, a fine, or both.
PetaPixel’s headline reference to “up to 10 years in prison” comes from adding the two separate maximum prison terms of three and seven years. It does not mean that a 10-year sentence has already been imposed or that such a sentence is inevitable. The man has been charged and the court process is still ongoing.
According to CNA, Ye Lin told the court through an interpreter that he intends to plead guilty and will not engage a lawyer. He is due to return to court on November 10.
A digital image with real-world consequences
The case shows that the problem with generative imagery is not limited to whether an output looks realistic. An insufficiently verified image can influence real-world decisions, deploy public resources and lead to activities being suspended for safety reasons.
For photographers, journalists and content creators, the provenance of an image is becoming increasingly important: who created it, where the original file came from, the context in which it appeared, and whether independent sources confirm what it is claimed to show.
What we think
The key point is not simply that someone could create a convincing crocodile image with AI. It is that a synthetic image was able to enter a real information chain as possible evidence.
The fact that PUB’s own AI detection tools did not provide a clear answer shows why verification cannot rely on a single technical tool. File provenance, context, primary sources and human verification become more important as synthetic imagery becomes more realistic.
Frequently asked questions
Was there really a crocodile at Pandan Reservoir?
Authorities confirmed that the photograph linked to this specific report was fake and ended the search operation. This case therefore does not constitute a confirmed crocodile sighting.
Was the image actually created with AI?
Yes. The Singapore Police Force says its investigations found that the accused had created an AI-generated image depicting a crocodile.
Why do some reports mention up to 10 years in prison?
The two separate charges carry maximum prison terms of three and seven years respectively. Adding those maximums explains the 10-year figure used in some headlines, but it is not an already imposed or guaranteed sentence.
Did AI detection tools identify the image?
Not with certainty. PUB said it used several tools, but the results were inconclusive.
