Is it real? A guide to identifying fake wildlife videos created with generative AI
Author’s Note: Despite the visual resemblance to the current GenAI This image by user sunny305 was published in 2021, before the AI boom, and likely represents the skills of a digital Photoshop artist.
The rapidly expanding world of generative AI is carving out a particularly popular niche that could have long-term damage to wildlife and the way humans interact with them. It’s a fake video of an animal.
Imagine this appearing in your feed. The fenced backyard is dark and nocturnal animals are out to explore. In the video, a group of rabbits on the edge of a trampoline examine the surface, move forward, notice the elasticity of the black mesh, and begin bouncing. Soon the whole group is jumping up and down with enthusiasm.
However, there are some problems. At the beginning of the video, there appear to be seven rabbits, but by the time they jump, there are six rabbits. It has blond hair, which is unusual for a wild rabbit, and the color pattern seems to have disappeared. In fact, one of the bouncing rabbits do disappear! What is this creepy, evil rabbit that changes shape and disappears from existence?
this video is not real;It was generated by a computer, now known as generative artificial intelligence. Generative AI is based on large language models that ingest large amounts of data and predict what to generate based on examples in the model.
The results of large-scale language model inference can be unexpected. Thus, a rabbit’s ears are reabsorbed into another rabbit’s fluffy butt. The model doesn’t know that this shouldn’t happen. Just know that if many rabbits are crowded together, ears that are not attached to the rabbit’s head can appear on the backs of other rabbits. A.I. “I don’t think about it” by itself— it reproduces what it sees from the given data. According to a 2024 study, AI systems Little awareness of object persistencea marker of infant cognitive development.
AI-generated wildlife videos receive millions of impressions and are posted to social media sites by thousands of social media “creators” every month. Some social media apps such as Facebook and Instagram We have posted the policy Some require users to label AI-generated images and videos as AI-generated, while others, such as X, do not currently require all users to do so.
But enforcement by social media companies is often patchy, with many users deceiving others by hiding the required “Made by AI” line at the end of a line of text or not following the rules at all. For example, this Instagram video of an eagle stepping onto wet concrete The artwork, watched by construction workers, has no clear “AI-made” identifier, and even though it was posted on Instagram, the caption makes it seem like it actually happened.
If you make a mistake in assuming that a large-scale language model is true, you can use the technical term for that mistake:hallucination”According to the study, hallucinations are inevitable According to a 2025 study by researchers at the University of Singapore, this is due to the nature of technology. Because large language models must act on their own predictions to continue, errors accumulate on each other and new generative AI models Show more errors Rather than less.
System overload
Although it may seem harmless, the misinformation spread by these types of posts causes more harm than it appears on the surface.
Cultural depictions of animals can sway public sentiment toward wild animals. The movie Jaws, released 51 years ago, contributed to long-standing negative perceptions of sharks and may have influenced the increase in shark killings. The proliferation of fake wildlife content on social media today can give people the wrong idea about animal behavior and how to safely interact with nature.
Most people who go hiking don’t expect to encounter a dangerous animal, but if they do, do they know what to do? And even if they do, what about a neighbor or an older uncle? If you saw a video of a grizzly bear licking a kitten, would you know that this is an unlikely event? Would a child who grew up watching fun fake animal videos be able to guess that this wasn’t real?
The cost of such generated AI videos could be as high as a life.
Such dangers are already evident in the occurrence of large numbers of organisms. AI-generated mushroom guide. As any hunter-gatherer knows, identifying edible food in the wild is already an important process, and no item is more important than mushrooms. There are several deadly mushroom species in North America that look almost identical to edible mushrooms and can only be distinguished by the most experienced mycologists. recent Alarming rise in mushroom poisoning in California Highlight the danger. Such toxic fungi can kill a person within hours, cause irreparable damage In your body if you are alive.
This type of information impacts more than just safety. A sense of awe and wonder at the world around us is an unexpected casualty of the proliferation of convincing fakes. It has been shown that creativity is impaired. Also as a side effect of using AI tools. why do you think the machine will do it for you?
read between the code
There are several common ways to determine if the video you’re watching is AI-generated.
video length: Most generative AI video sources can only generate 30 seconds of video at a time, with little consistency. LLMs often don’t have a lot of memory and will create a new video every time the software prompts you. Therefore, the model slightly or dramatically changes the appearance of the characters in the video from prompt to prompt or scene to scene. Some video prompters have come up with complex workarounds to maintain consistency, but for the time being, most accounts that intentionally post generated AI videos won’t go that far.
consider the source: AI accounts often post multiple versions of similar videos in hopes of gaining views and likes that will lead to monetization. Does the source have many similar videos, or does the video seem intentional? Is the source a new account, or one that appears unresponsive? Many of these types of accounts across platforms profit from clickbait and misinformation.
visual clues: A few years ago, there was a simple way to spot fake images by counting the fingers on a hand in a photo that appeared to be AI. Large language models are becoming more complex, but glitches can still occur, like in the rabbit and trampoline video. Mistakes can be more subtle, so ask yourself:
- Do the colors, sizes, and movements of the animals look natural?
- What about settings?
- Do straight lines on teeth, bricks, tiles, walls, etc. become blurry or disappear?
- Can you tell where the light in the video is coming from (sun, lamp, etc.)?
- Does the direction of the light change? Does the shadow move on its own?
- Does video quality matter? One of the reasons the bunny video fools us is because we expect it to look like security footage and be low resolution.
Media date: If the video or image was posted before 2022, it’s very likely to be authentic. Until generative AI tools became easily accessible to the general public, creating convincing fake images required significantly more work on the part of the individual.
Reverse image search: There are several versions of this on different search engines, but the idea is that you can enter an image and it will show you all the sites where that image can be found. This can help you determine if an image is real or possibly a fake that resurfaces every few years, and can help you find what may have been posted in the first place.
Video content: Ask yourself if the behavior makes sense. If this actually happened in nature, other videos and articles could be made about it. Consult expert sites and authoritative forum posts to determine what actually happens. Try to use verified sources whenever possible, as search engines can show you sites and pages that exist just for clicks. Unfortunately, AI detectors are not always accurate and seem to become less accurate over time.
trusted sources
The best way to determine the reality of what you’re seeing is to have a place for expert opinion.
Below are some resources commonly used (even by wildlife biologists) to identify wildlife and learn about their behavior.
- i naturalist: iNaturalist is a community science app and website where anyone can upload photos of anything from animals to plants to fungi and rely on real people to help identify them, often very quickly. iNaturalist also offers the ability to view hundreds of photos of common wildlife, allowing ID seekers to see unusual characteristics that may occur in a species. Have you seen a screech owl molting? I’ll also include a photo for comparison!
- merlin: Merlin is a bird-specific app and website similar to iNaturalist, but with sample bird calls that you can use to confirm your identity. It is a favorite of avid birders and has a reputation.
- Maryland Biodiversity Project: Since 2012, MBP has been on a mission to catalog species found at state borders, and we’re thorough in our efforts.
- hell mapper Similar to iNaturalist, but for amphibians and reptiles.
- Local wildlife groups: Your greatest asset in determining whether you are getting the correct information is the experience of those who are familiar with local wildlife. Compiling a group’s collective knowledge often takes precedence over doing your own research. For example, a group of birders with varying degrees of experience may have more than a century of combined experience. Humans have always shared collective knowledge, but times like this show how important that practice is.
Consulting these sources to obtain an ID will take more time than asking your AI assistant, but it’s well worth the effort and you’ll learn something new in the process.
If you see something you’re not sure about, be wary and don’t share it if you think it might be AI. Sharing reinforces and spreads misinformation and encourages the creation of new posts. Social media and AI companies make a ton of money off of our usage, even when we don’t ask for it (and many don’t). It’s easy to blame others for sharing, but remember that they face the same uncertainties as we do.
The real world, and the real wonders it contains, are worth fighting for. Appreciate real nature when you are surrounded by man-made things. There’s great wildlife material out there without resorting to crude imitations or impossible rabbits.
