Clips about current events aren’t the only things that go viral. Many feature dangerous interactions between humans and animals. For example, several videos, including one with a visible Sora watermark, show dogs racing to save a baby snatched into the air by an eagle.
These videos bear a striking resemblance to fake videos that circulated long before generative AI captured our attention online.
“Golden Eagle Snatches Kid” was posted to YouTube in 2012 and has been viewed more than 47 million times. Now, the video by four 3D animation and design students at Montreal’s School of Digital Arts feels like a relic from a time before internet users could generate viral content with the click of a button.
“Back in the day, there weren’t many videos like this,” says Félix Marquis Poulain, a 3D character modeler who worked on the video as a student at the Université du Québec Chicoutimi’s Faculty of Digital Arts, Animation and Design, or École NAD-UQAC.
This video is the result of a class assignment in which students were asked to create an online hoax. With 100,000 views to earn an A-plus, Marquis Poulain and his classmates brainstormed a video that would be engaging enough to grab people’s attention. At the time, videos of babies and animals were popular on YouTube, so they thought the combination of baby and eagle was a sure bet.
“We took a story that was ending on a positive note, but it created this danger zone and it was believable in a way,” Marquis Poulain said.
The students spent approximately seven weeks working on the assignment, which required hundreds of hours of pre-production, filming, and post-production. The result was an organic viral video that received 3 million views on its first day, Marquis Poulain said.
He said he viewed the imitation AI baby and eagle video as “junk.”
“It’s like plastic waste,” he said. “[Generative AI]is an interesting new technology and a great tool, but at some point it gets into this production cycle and you can’t get rid of it, it just accumulates and accumulates and accumulates.”
Growing distrust online
The release of the Sora 2 app immediately sparked controversy as some users chose to create fake videos featuring the lives and deaths of celebrities and public figures.
The families of late human rights activist Malcolm X and actor Robin Williams have spoken out after a Sola video featuring them went viral.
At the request of Martin Luther King Jr.’s estate, OpenAI blocked Sola from producing a video featuring King’s likeness. Some users had created videos depicting Dr. King making rude and racist remarks, which OpenAI acknowledged were “disrespectful.”
Sora has safeguards in place to allow users to control whether their likeness can be used by others. It also blocks attempts to generate harmful content and includes watermarks and metadata that allow viewers to track videos back to Sora.
“Where there’s a guardrail, there’s a way around it,” said Mike Zaiko, a sociologist and associate professor at the University of British Columbia’s Okanagan campus. For example, it is possible and relatively easy to remove watermarks from AI tools.
Zaiko said generative AI tools are creating a sense of distrust among internet users who can no longer believe what they see.
“One of the best skills to develop is good old-fashioned digital literacy… using search engines and video image search tools to try to see if these things actually happened,” he said.
Platforms like Sora typically tout how groundbreaking their tools are, but “it’s been clear for some time that video generation is kind of the next frontier,” he added.
“I think this is an important step forward, and video generation platforms will continue to develop. But fundamentally, like anything else (generative AI), they reuse, remix, regurgitate cultural products that already exist. So they’re not great for creativity yet, but they’re great for remaking existing parts of human culture,” he said.
Impact on the industry
Marquis Poulain said he hasn’t yet seen generative AI replace many jobs in his industry, but “I can see it coming.”
He wants more regulation and takes issue with the fact that many AI models are trained on artists’ work without their consent. But he said there could be a place for AI in the visual effects industry by replacing the more tedious technical aspects of the work.
“There’s a lot of debate in the industry because some people are already interested in using it, and others are saying, ‘No, we should never mess with this. It’s going to destroy us, it’s going to replace us,'” he said.
How to find videos generated by AI
1. Find the watermark
Text-to-video generative models like Sora often include visible watermarks in the video. However, these can be deleted. In this fake video The work depicts the arrest of a worker by US immigration authorities, and the author placed an icon where Sora’s watermark would go. Some platforms, such as TikTok and YouTube, may state that the content is AI-generated in the video description.
2. Look for what is missing or wrong
Missing body parts, nonsensical text, and other “unusual” features can help identify AI content. For example, this AI-generated video of the CN Tower fire included cars without license plates and an unnatural, shiny appearance. another video Mossad, which is supposed to indicate the destruction of Israel’s intelligence agency, misspelled the agency’s name.
3. Hear unnatural sounds
Watch for unnatural or awkward speech, mispronunciation of common words, awkward pauses, or a monotone tone. Search for phrases from videos using a search engine. A Canadian Press search using a passage from a video featuring former public health officer Dr. Theresa Tam found an identical scripted video featuring Britain’s chief medical officer of health. Both videos were fake.
4. Fact-checking information
If a video makes a particular claim, use a search engine to see if legitimate sources such as government reports, peer-reviewed studies, trusted news outlets or fact-checking websites are talking about the same thing. If a specific person is mentioned in the claim, you can debunk false information by asking that person directly.
5. Find the original source
Some online tools can help you check the authenticity of your video. Taking a screenshot of the video and running it through a reverse image search tool like Google Lens will help you find the original source and confirm if it’s real.
This report was first published by Canadian Press November 21, 2025.
Marissa Barney, Canadian Press
