These images have been circulating on social media and video platforms such as YouTube and depict a large creature (that appears to be roughly the size of a large dog) with gills and webbed limbs in a fishing vessel. Social media posts have claimed that this is a recent discovery of an animal that has “never been seen before.”While it is true that new, previously unknown deep-sea creatures are still regularly being discovered, a closer look at the images raises several red flags. For instance, the number of gills and placement of coloured patches on the creature differ across images. In one image, the creature only appears to have one large gill on either side of its head, while in another, the “same” creature has over 5 gills on each side (see differences circled in the image below).These inconsistencies are clear indicators of an AI generated image. Coupled with the lack of any other pictures that can verify the creature’s existence on any reputable news sources, we flagged this image as very likely a fabrication. Further, the creature closely resembles an axolotl – which is type of salamander that typically does not exceed 30cm in length – suggesting that the existing creature was likely referenced by the AI to generate the claim images.Upon reverse searching the claim images, we found that they were first posted in July this year by a Facebook user named Astral Internum Productions. The page describes itself as doing “AI assisted World Building and Storyboarding” with art made with “a combination of AI, Sketches, and Photoshop.”As an article by AFP Factcheck also points out, the page owner is clear in the comments about their content being AI and was not part of the subsequent spread of the images as an “unknown catch by Papuan fishermen.”It therefore appears that not a single element of the claim is accurate, as the images were created by an AI content page that made no claims of authenticity. We give this claim a rating of false.
Building resistance to AI generated images is more important than ever given the growing ubiquity of such content on the internet. Being able to identify inconsistencies and red flags even when it comes to light-hearted, human-interest type posts should be part of our toolkit in maintaining awareness of (and resistance against) mis/disinformation.