"If a record number of fatal bear attacks wasn't terrifying enough, experts say a torrent of AI-generated videos in Japan purporting to show people in close encounters with the animals is only adding to public anxiety and could put people at greater risk. While headlines about real attacks and disruption appear on a regular basis, monitors of online content are warning social media users not to be taken in by realistic videos on platforms such as TikTok of the animals attacking or interacting with humans."
"And they can easily outrun humans. Experts hope most people will be able to distinguish real from fake content, but the sheer number of clips could exacerbate fears that Japan is in the grip of an outbreak of ursine terror. There is concern too that the videos could undermine government messaging on what to do during an encounter with a bear, either in the wild or, increasingly, in towns and other built-up areas, where they search for food."
A record number of fatal bear attacks in Japan is accompanied by a surge of AI-generated videos that purport to show close human–bear encounters. Realistic clips on platforms like TikTok include bears destroying solar panels, carrying off a pet dog, a newsreader reporting sightings, and staged scenes of people feeding or fighting bears. Some clips were produced using OpenAI's video generator and bear a Sora watermark; a TikTok search found roughly 60% of sampled clips were fake. The volume and realism of the videos risk inflaming public fear, confusing viewers, and undermining official guidance on safe responses during bear encounters.
Read at www.theguardian.com
Unable to calculate read time
Collection
[
|
...
]