In recent weeks, social networks have been abuzz with short videos of impressive realism, often produced through OpenAI’s Sora2 artificial intelligence tool. Some mimic security camera footage, others resemble clips from animal documentaries — but all have one thing in common: they’re completely AI-generated.
Although OpenAI’s Sora 2 software is officially available only in the United States and Canadathese AI-generated videos already have take social media by storm around the worldwrites the French-speaking Swiss public television RTS.
These contents raise serious questions about the trustworthiness of images on the internet. Besides, the OpenAI has already been accused of allowing the spread of videos using the image of celebrities — even deceased — without permission or respect for copyright.
“Our brains are easily tricked”
Speaking on an RTS radio show, psychologist and psychotherapist Nils Weber explained why these fake images seem so convincing: “Our brain is not difficult to trick; it is relatively easy to make it believe things,” he noted.
He even compared this phenomenon to the first cinematic illusions created by Georges Méliès at the beginning of the 20th century: “Suddenly, people wondered: what is happening?”. Georges Méliès was a French magician and director, a pioneer of special effects in cinema. With works such as A Trip to the Moon (1902), he used tricks, duplications and innovative editing to create the illusion of the fantastical on screen.
According to the expert, these videos have only one purpose: to get our attention. The platforms themselves promote them more, because they cause more interaction.
Learning to doubt and “decode”
Niels Weber believes that now “we have to question everything we see on social networks”. For him, media education must be strengthened, both for children and adults: “We must learn to recognize inconsistencies, objects that move unnaturally or behaviors that seem unrealistic”.
The psychologist cited as an example a fake “security camera” video showing a bear going into a toilet: “The real question is why would there be a camera in a toilet?”
In general, we should always give special attention to the details of a video or image: sometimes, distortions or unnatural movements may be an indication that the content has been artificially generated.
Additionally, Sora 2 normally adds one watermark on the videos it creates. However, many users remove it. For this reason, even more care is needed: in some cases, a small blurry square appearing on images is an indication that Sora 2’s watermark has been removed.
There are social media accounts that offer tips on how to identify such fake images and videos, while some online videos explain techniques for spotting AI-generated content.
Removal from social networks?
RTS believes that the excessive dissemination of such content may tire users over time. At the same time, in Europe OpenAI will have to comply with strict regulations, but similar applications are expected to appear quickly without adequate safeguards, increasing the risk of disinformation.
To avoid fake news, the solution might be to stay away from social networks. Niels Weber, however, emphasizes that this will not necessarily be a conscious choice but a natural evolution: “Users will get tired of constantly having to question what they see and will eventually realize that these platforms no longer offer them what they were looking for in the first place.”
