AI can now fake the videos we trust most. How to tell the difference-and how newsrooms can respond
Briefly

AI can now fake the videos we trust most. How to tell the difference-and how newsrooms can respond
"It was an amusing mix-up, but it was also a warning. AI-generated video tools have moved far beyond producing surreal or obviously manipulated clips. They are now convincingly imitating the formats we instinctively trust most: CCTV, dashcams, police bodycams, wildlife cameras, and handheld eyewitness footage. These are the clips that shape public understanding during protests, disasters, violence, and emergencies. And the fake ones are becoming indistinguishable from the real thing."
"At Storyful, we verify thousands of real-world videos for newsrooms and brands worldwide. This year we ran a test: we fed real breaking-news headlines from our own platform into one of the newest AI video models. In seconds, we got clips that mimicked the texture and perspective of eyewitness reporting. Not glossy AI experiments, news-like footage that could plausibly land in a newsroom inbox during a breaking story."
"In seconds, we got clips that mimicked the texture and perspective of eyewitness reporting. Not glossy AI experiments, news-like footage that could plausibly land in a newsroom inbox during a breaking story. Side by side with the original real clips, even trained journalists needed to slow down and scrutinize the details. Consider this example, inspired by a verified authentic video posted to social media in the wake of heavy monsoon rains in India:"
AI-generated video tools can now produce highly realistic footage that mimics CCTV, dashcams, police bodycams, wildlife cameras, and handheld eyewitness perspectives. A verified security-camera clip of a coyote on a trampoline went viral alongside near-identical AI-generated animal clips shared as authentic. Storyful verified thousands of videos and tested an AI model by supplying breaking-news headlines; the model produced news-like clips in seconds that matched eyewitness texture and perspective. Side-by-side comparisons showed even trained journalists needed extra scrutiny to tell real from fake. These capabilities risk undermining public understanding during protests, disasters, violence, and emergencies.
Read at Fast Company
Unable to calculate read time
[
|
]