AI-generated photos of ‘giant marine creatures’ falsely linked to Fukushima’s wastewater discharge

 

 Screenshots of the viral photos.

 

News Brief

Multiple photos shared Aug. 28, 2023, on Facebook, claim to show people posing with some giant marine creatures. Same photos were also circulating here and here. In these photos, the size of the marine creatures appears unusually large. The Chinese-language captions translate as, “the world has changed after Japan’s discharge of radioactive wastewater,” “various mutated animals living in the sea has started appearing in different areas of the South China Sea due to Japan’s discharge of radioactive water,” and “fishermen on Xiaohongshu say that radioactive contamination has arrived.” Xiaohongshu is a Chinese social media platform, sometimes referred to as the “Chinese Instagram.” These captions suggest that the giant creatures in the photos have undergone genetic mutations due to the radioactive contamination of the wastewater from Fukushima’s nuclear plant.

As of the issuance of this report, the Facebook post had been shared 129 times, and it had received 149 comments and 401 likes or reactions.

 

Fact-checking

Multiple inconsistencies have been discovered in the viral photos. For instance, one person had two heads, and the hand shapes of two individuals appeared unnatural.

 

Numerous inconsistencies were found in the viral photos.

 

Based on the Xiaohongshu account superimposed on the bottom right of the photos, a Xiaohongshu post with five exactly same photos was found. The caption, hashtagged “AI Drawing,” claimed, “waiting for months, Shanwei finally resumed fishing for seafood!!” The post was last edited on Aug. 26, 2023, two days after Japan began discharging radioactive wastewater.

 

The Xiaohongshu account is shown in the bottom right corner of the viral photo.

 

The Xiaohongshu post, hashtagged “AI Drawing,” displayed the same photos as the viral ones.

 

Some netizens commented on the Xiaohongshu post that the giant marine creatures in the photos were related to Japan’s radioactive wastewater. In response, the post author clarified, “I apologize if it made anyone uncomfortable! But I want to emphasize again that these are AI-generated photos! Don’t take them too seriously!” The author also claims that the images were reposted from elsewhere. Therefore, we can conclude that the viral photos are not genuine.

The AI-generated content detection program Hive Moderation was used to analyze the five photos. The results showed a 99.9% likelihood that the photos were generated by AI, with a strong possibility of being created using the AI image generator Midjourney.

By entering a prompt (a textual description of the image) in the chat box of the Midjourney bot on Discord, users can create impressive and photorealistic artwork. Multiple fact checks published by HKBU Fact Check have debunked false information related to images generated by Midjourney, such as the stories of Trump’s arrest and frost flower.

 

The results from the AI-generated content detection program Hive Moderation.

 

To learn more about how to identify AI-generated images, please refer to the article published by HKBU Fact Check, which introduces tools and techniques for detecting such images.