A Japanese news agency’s decision to withdraw a wildlife photo enhanced by artificial intelligence (AI) has reignited debate over the technology’s fast-spreading role in conservation, a field now worth billions as AI reshapes how the natural world is documented and protected.
Kyodo News said it pulled the image, released on 20 October, after discovering that a collaborator for a local nonprofit had used a generative AI tool to “enhance” footage of a raccoon dog, locally known as tanuki, carrying a baby sea turtle on Yakushima Island in southern Japan.
(Top) Image of a raccoon dog captured by a surveillance camera; (Bottom) image of the raccoon dog processed using generative AI. Image: Yakushima Umigame-kan
Yakushima Island’s beaches are among the most critical in the North Pacific for the endangered Loggerhead sea turtle and Green sea turtle with between 30 and 40 per cent of all loggerheads nesting in Japan doing so on Nagata Beach alone.
The edit, meant to clarify the image, altered the animal’s posture and the hatchling’s position, prompting the agency to call it “inappropriate for reportage”.
“It was deemed inappropriate as a news photograph that should accurately convey the facts,” said Kyodo News in a correction.
While Japan’s Ministry of the Environment confirmed that tanuki predation of turtle hatchlings does occur on the island, the manipulation has renewed scrutiny of how AI tools can distort evidence in scientific and journalistic work.
Rise in AI use
Market research firm Allied Market Research estimates that the global AI-in-wildlife-conservation market was valued at US$1.8 billion in 2023 and could surge to US $16.5 billion by 2032, showing the rush by non-governmental organisations (NGOs), tech firms and governments to automate biodiversity tracking.
Projects such as Wildlife Insights, a global database backed by Google and Conservation International, for instance, now rely on AI trained on over 35 million labelled camera-trap images across nearly 1,300 species.
The system can sort millions of photos in minutes, a task that once took humans months, helping conservationists monitor animal populations at scale.
World Wildlife Fund (WWF) says similar tools are being used to predict deforestation and detect illegal logging, while start-ups in Africa and Asia deploy acoustic AI to identify gunshots or chainsaws in protected areas.
Private companies are also capitalising on the boom, offering AI-powered image-recognition platforms and data-analysis services to NGOs and government agencies, which could reduce labour costs, improve accuracy and create new commercial ecosystems around conservation data, according to analysts.
However, critics warn that the same technologies transforming conservation could also undermine it.
According to a 2025 study published in Conservation Biology, AI-generated wildlife images and videos may blur the line between reality and fabrication, threatening public trust in conservation media.
“On social media, AI-generated, realistic yet misleading portrayals of wildlife are reshaping public attitudes toward biodiversity,” the authors of the study said. “Generative AI allows for effortless creation of entirely fictitious species, behaviours, and ecological interactions, further distancing the public from real-world conservation challenges.”
Another analysis published in Trends in Ecology & Evolution in February described emerging risks of what researchers call “AI colonialism” – the growing dependence of conservation projects in the Global South on artificial intelligence systems developed in wealthier nations.
The authors warned that many of these machine-learning tools are trained on datasets, priorities and definitions designed in the Global North, yet deployed in biodiversity-rich regions such as Southeast Asia, Africa and Latin America with little local oversight or capacity-building.
This imbalance, they said, can reinforce existing power dynamics in global conservation, shaping which species are monitored, how data are interpreted and who controls access to environmental information.
Fragmented regulations
Efforts to regulate the use of AI in conservation remain fragmented and slow.
While governments and international bodies such as the European Union and the United Nations have begun crafting broad AI governance frameworks, none specifically address environmental or wildlife applications.
Most conservation projects now operate in a legal grey zone, relying on internal ethics policies rather than enforceable standards. Researchers writing in Conservation Letters say this gap leaves major blind spots around data privacy, algorithmic bias, and the potential misuse of AI-generated imagery in public communications.
In developing countries, where much biodiversity research takes place, regulation is even further behind. A Nigerian legal review on AI and wildlife protection noted that deploying machine-learning tools without oversight could “reinforce existing compliance gaps” in forest and species protection.
Similar warnings have been raised in Asia and Africa, where imported AI systems often lack local consent mechanisms or clear accountability structures.
