The Impact of AI-Generated Media on Animal Content
In recent years, the surge of AI-generated videos and images has been undermining one of the most cherished sources of joy and compassion found online: animal content. Whether it involves sailors rescuing polar bears, calves separated from their mothers, or lost pets miraculously reunited with their owners, many individuals have reached a point where they can no longer trust what they see in their feeds. The rapid proliferation of deepfakes and low-quality AI-generated content has led to widespread frustration among viewers, creators, and animal welfare organisations, compelling everyone to scrutinise pixels for signs of technical flaws or other indicators of AI intervention.
Mibbby Butler experienced a wave of joy one April day when she received a text message stating that her lost cat, Brooklyn, had been found. Allegedly, her flatmate’s boyfriend had taken Brooklyn and abandoned him the previous day, roughly 30 minutes away from Los Angeles. Butler promptly posted a ‘missing’ notice on social media and began searching the area when a stranger sent her a photo. In this image, Brooklyn was seated on a kitchen countertop, being embraced by a young woman.
“My little one is safe!” Butler exclaimed. However, to reclaim her pet, she was required to pay upfront for Brooklyn’s supposed temporary care. It was at this moment that Butler suspected she was being scammed and began examining the photo more closely. Brooklyn posed in the same manner as in the missing poster, and both images featured a bottle of Torani syrup and a microwave in the background. However, in the new photo, the text on the label was distorted, a classic sign of AI generation. Butler refrained from making any payment, stating, “I had no idea people would scam others with lost pets,” and chose not to report the incident to the police, unsure if any crime had been committed.
Four months later, Brooklyn remains missing.
The Challenges for Authentic Content Creators
The overwhelming tide of fakes has also made life more difficult for creators of genuine content. The non-profit organisation We Animals has published works from 175 photojournalists documenting alleged cases of animal cruelty and other abuses in farms, circuses, and scientific laboratories. The material can be distressing, which often breeds skepticism, as noted by Eva von Jagow, the organisation’s marketing director. For instance, drone footage shared by We Animals, depicting rows upon rows of cages purportedly used to house calves separated from their mothers in an Arizona dairy farm, appears almost too perfect to be real.
In the past, some critics claimed that We Animals’ work was staged or photoshopped. It is no surprise, then, that viewers sometimes accuse the organisation of employing AI, despite its strict prohibition against the use of such technology by its photographers. “How do we know it’s not AI?” commented an Instagram user earlier this month regarding a video from the dairy farm.
To reassure the public of the authenticity of their content, We Animals plans to share behind-the-scenes videos and provide more details about the processes their staff follows to verify received material. They will also adopt technology that incorporates provenance and editing history into the digital files of photographs and clips. Victoria de Martigny, the group’s visual content director, emphasises the importance of maintaining trust, as losing it could lead to widespread questioning of all their work, a situation they wish to avoid.
The Consequences of AI-Generated Animal Content
Genuine and evocative images often resonate deeply with viewers, serving as a crucial avenue for conservation and welfare projects to secure donations. However, for those who disseminate low-quality, AI-generated content, manipulated images can be an easy route to accumulate ‘likes’ and advertising revenue. Oscar Horta, a philosopher and prominent animal rights activist who has recently contributed to a short film exploring the potential impacts of AI on wildlife, notes that fakes are much cheaper to produce and appear to be crowding out real videos on social media and search results.
Horta expresses concern over the absurd clips of animals he has encountered online lately, such as those featuring wild animals being rescued from fires and floods. He worries that this AI-generated content may lead people to question legitimate rescue efforts, potentially hindering future fundraising endeavours. This is particularly troubling in an era where extreme weather events are becoming increasingly common.
Furthermore, the proliferation of false content could inadvertently encourage harmful behaviour towards animals. “There are deepfakes of polar bears drowning, with people in boats attempting to rescue them,” states Horta, describing these scenes as “ridiculous” and “poorly representative of what it means to help animals.”
Addressing the Challenge of Misinformation
Researchers have begun proposing strategies to curtail the creation and dissemination of AI-generated misinformation about animals. This month, Jeff Sebo, director of the Centre for Mind, Ethics, and Policy at New York University, urged AI developers to embed guidelines in their models that discourage generating potentially harmful responses toward animals. The aim is to “emphasise the importance of evidence-based reasoning” concerning the possible suffering of specific creatures while avoiding a tone that is overly moralistic or dismissive of reasonable user requests.
New legislation in California and the EU mandates that popular AI image generators incorporate invisible tags indicating that the content has been AI-generated, a requirement that also extends to social media platforms. Increasing public awareness about AI and broader access to verification tools could help restore trust. Features from ChatGPT, Gemini, and Meta AI enable users to identify whether an image has been generated by the respective chatbot. However, most individuals lack the time to upload every animal-related piece of content they encounter to these tools, and limitations exist, such as caps on the number of images that can be verified.
Integrating verification tools directly within messaging apps and web browsers, with default activation, could enhance their usefulness significantly. Butler’s distress might have been alleviated had her iPhone automatically alerted her that the photo of Brooklyn sent by the initial scammer could be false. She reports receiving a deepfake about once a month from someone claiming to have found her cat. Tests conducted by WIRED confirmed that ChatGPT generated at least one of these images, depicting a supposed Brooklyn eating in front of a door. Using a 29-word prompt along with the original missing poster of Brooklyn, WIRED was able to create an almost identical image to that of the scammer.
Butler feels trapped amid the AI-generated cat videos that flood her online experience. When she stumbles upon one in her social media feeds, she blocks the user who posted it, yet new accounts continue to emerge. She now spends more time in a Facebook group comprising nearly 300,000 members who oppose AI, which has become a new sanctuary for her. Here, she can appreciate art that she knows was created by humans while she waits for Brooklyn’s return.