But Vermilya and other experts are more worried about negative impacts. Some research suggests that just seeing lots of images of a certain species—even in the form of children’s toys or commercials—can give the impression that these animals are more abundant than they actually are. That’s a problem when, as is the case with many of the AI-generated bird videos, social media content depicts endangered animals—creatures in dire need of conservation attention—living plentiful lives. Gosler recalls a video showing a threatened tropical bird using its wings as an umbrella to shield its young from the rain, evoking a warm, fuzzy sense that all is well with that species. “I think that’s probably one of the biggest issues—that it gives the impression that actually the environmental messages are less serious than they actually are,” Gosler said. At its extreme, this could lead to fewer donations and volunteer efforts toward conserving species.

Another concern is that people will believe the inaccurate details such videos contain. Even the most ridiculous concoctions, such as a Facebook video featuring a man reviving a dead robin by dunking the bird in a bucket of water, attract comments suggesting the images are real; in the case of the robin video, Gosler recalls, viewers praised the man’s kindness. (Needless to say, “You can’t revive dead robins by immersing them in water,” Gosler said.) People are generally more prone to believing misinformation when they see it in the form of videos, according to misinformation and generative AI expert Jevin West of the University of Washington, who co-founded the Center for an Informed Public. “Even if you almost know they’re fake, it’s hard to dislodge yourself from them not being real,” West said.

People in many urban Western societies may be more easily duped because they already know so little about wildlife. In a yet-unpublished survey of roughly 1,000 Spanish preschool and elementary school students, biologists José Guerrero Casado and Tamara Murillo Jiménez at Spain’s Universidad de Córdoba found that the children had surprisingly poor knowledge about animals, and especially native species; many didn’t know that zebras and tigers are not native to Spain or that wild rabbits don’t actually eat carrots. And in 2021, Gosler found similarly poor ecological knowledge in a survey of 149 first-year biology students at Oxford: only 55.7% could name five species of local British birds. “They’re so disconnected from nature,” Gosler lamented. 

AI-generated imagery could further distort people’s knowledge of animals in numerous ways. Vermilya, for instance, is worried about the many AI-generated videos showing animals with humanlike behaviors and feelings. Studies on children who read books featuring anthropomorphized animal characters suggest that, while such depictions can boost empathy and compassion toward animals, they can also lead young people to believe that animals are more humanlike than they actually are. Portraying animals such as bears as cuddly and friendly creatures could even be dangerous, since people may be more inclined to seek out such animals in the wild, Vermilya cautions. “I’m near the Rocky Mountains, and there’s a … long history of people going off into the mountains and wanting to interact with wild animals and getting mauled,” she said. 

Conversely, AI-generated imagery of predatory animals violently attacking humans or livestock, or evil-seeming snakes hunting a bird’s young chicks, for instance, could create unnecessary levels of fear toward a species and lead people to believe certain animals are less deserving of conservation attention, even if they’re endangered. “If you interpret that the snake is a bad animal—as a devil—and the chicks and the bird are good animals, [then it seems that] we have to conserve the bird but not the snake,” said Casado, who, together with Jiménez, wrote an article published in September about the risks of AI-generated wildlife videos.

Misrepresenting animals’ behaviors or environments can distort people’s understanding of where animals truly belong and how important they are to their ecosystems. “That can lead to, say, the decision-making around, well, ‘How do I think about that animal? Do I support organizations that help that animal? Do I go and visit that animal in that space? Do I eat that animal?’” Vermilya said. A video showing a capybara, a South American rodent, in a bathtub, for instance, could boost people’s interest in keeping such animals as pets and fuel the illegal trade of such species, Jiménez says. 

Images of animals that don’t exist at all, like hybrids of hippos and lizards, could cause confusion among children and young people whose critical thinking skills haven’t yet developed. “This is a real problem now,” Jiménez said of young people spending much of their time online and indoors, far from natural environments. “They think the things that they’re seeing in their phones … is reality.”  

There’s also risk to people finding out certain AI-generated imagery is fake, West says, as doing so can make it harder to trust accurate depictions, including photos and videos taken by professional wildlife photographers who may spend months working to get a shot of a rare species or by scientists to document an animal-related scientific discovery. “What if we did see a hummingbird in some interesting situation, and we’ll just think, ‘Oh, it’s a deep fake,’ when, in fact, it might actually be real?” West asked. “Then we’ve lost the opportunity to be excited about biology.”