Japan is confronting a compounding crisis as artificial intelligence-generated videos depicting bear attacks flood social media platforms, threatening to amplify public panic during a genuine surge in wildlife encounters. The phenomenon underscores a broader challenge facing governments worldwide: maintaining public trust and safety guidance when AI-generated disinformation can spread faster than verified information.
A search by the Yomiuri Shimbun newspaper of bear-related content on TikTok revealed that approximately 60 percent of sampled clips were artificially generated using OpenAI’s Sora video synthesis tool. While some creators watermarked their content and labeled it as fabricated, others deliberately obscured the artificial origin, crafting videos with sufficient visual fidelity to deceive casual viewers. The clips portrayed scenes of bears destroying solar panels, attacking pets, entering convenience stores, and—most dangerously—being fed by residents in violation of wildlife management protocols.
Japan is experiencing an unprecedented wildlife crisis independent of any AI-driven distortion. The environment ministry confirmed 13 fatal bear attacks in 2025, more than double the previous annual record, with over 100 additional injuries reported. Between April and September, authorities documented roughly 20,700 bear sightings nationwide—approximately 7,000 more incidents than the same period in 2024. Black bears in Japan typically weigh up to 130 kilograms, while Ussuri brown bears inhabiting Hokkaido can reach 400 kilograms, making them capable of outrunning and overpowering humans with ease.
Environmental Drivers Behind Wildlife Displacement
Wildlife experts attribute the spike to poor yields of acorns and beechnuts, which comprise the bulk of bears’ natural diet, forcing animals into populated areas in search of food. This ecological pressure intersects with decades of rural depopulation in northern Japan, which has eroded natural boundaries between forests and residential zones. As bears venture further into towns and villages, infrastructure operators have begun scaling back services. Japan Post suspended mail collections and deliveries in parts of Akita prefecture, where members of Japan’s Self-Defence Forces have been deployed alongside licensed hunters to trap and euthanize bears. Telecommunications firm NTT East announced plans to revise safety protocols for engineers maintaining mountain communications infrastructure, while logistics company Yamato warned of potential service suspensions if worker safety cannot be assured.
The proliferation of AI-generated content compounds this legitimate threat by potentially undermining government messaging on wildlife safety. Experts and local officials have expressed particular alarm over fake videos depicting residents feeding bears—a behavior that wildlife managers warn conditions animals to lose their natural fear of humans, dramatically increasing attack risk. In one fabricated video, a fake news broadcast reported a bear entering a convenience store in Noshiro, Akita prefecture, prompting local officials to issue a public denial and call for residents to disregard unverified social media content. A second fabricated clip showed pedestrians fleeing from a bear on an Ishikawa prefecture street, similarly refuted by regional authorities.
Shinsuke Koike, a professor at Tokyo University of Agriculture and Technology, told the Yomiuri that feeding bears represents an existential threat to public safety. “Feeding a bear is extremely dangerous, and could result in them no longer fearing humans,” Koike stated, emphasizing that such videos—whether real or fabricated—normalize behavior that directly contradicts official guidance.
Disinformation as Public Health Threat
The convergence of genuine wildlife crisis and AI-amplified misinformation illustrates a widening vulnerability in public risk communication. When AI-generated content can realistically simulate news broadcasts, expert commentary, and eyewitness video, citizens face heightened difficulty distinguishing authoritative guidance from fabrication. Several fake videos included geographically specific details—referencing actual locations where bear sightings made headlines—increasing their perceived credibility among viewers unfamiliar with the incidents.
The timing of the AI video surge is not coincidental. As media coverage of real attacks intensified, social media users capitalized on heightened public attention by generating and sharing synthetic content designed to accumulate engagement metrics. This incentive structure—where algorithmic promotion rewards content generating clicks regardless of veracity—creates conditions for disinformation to proliferate during genuine emergencies.
Japan’s experience reflects a global pattern. As international bodies including the United Nations have documented, AI-generated content poses mounting risks to public health responses and emergency management. The challenge intensifies when synthetic media can undermine trust in legitimate institutional warnings, potentially leading populations to disregard verified safety protocols.
Regional authorities have begun issuing coordinated denials and appeals for citizens to rely exclusively on official government sources for bear safety information. However, the volume and visual sophistication of fabricated content suggests traditional press releases and local media briefings may prove insufficient to contain the spread of misinformation at scale.
Japan’s dual crisis—ecological displacement of wildlife colliding with disinformation saturation—will likely serve as a test case for how democratic governments manage public communication during emergencies in an age of synthetic media. The outcome may inform policy responses across other nations confronting wildlife management challenges, agricultural pressures, or other crises vulnerable to AI-assisted disinformation campaigns.
