A video that appears to show a British soldier yelling at police about arresting “people for words” has been created using artificial intelligence.
The clip has been widely shared on social media, with X reposting it 16,000 times.
However, experts in identifying AI-generated content said there were signs that the video was not authentic and was likely created using AI. When Full Fact contacted OpenAI, which owns the AI video generation software Sora, they said they had taken action and banned the video's original creator (not the re-sharer) from the platform.
In the video, a soldier appears to tell a police officer: “I sacrificed my life to protect the freedom and freedom of speech in this country. And yet you turn a blind eye to the crimes committed by the invaders and arrest people for words. You are a disgrace to the uniform and an insult to me.”
Dr. Shiwei Liu, a digital media forensics expert at the State University of New York at Buffalo, told Full Fact after analyzing the clip that it was “likely to have been generated by AI.”
He said its 15-second length suggests it may have been created with Sora 2, an AI video generator created by OpenAI, and cropped to remove any visible watermarks.
“There are unnatural facial expressions, a complete lack of blinking and obvious mistakes with the name tag on the uniform being repeated,” he added.
“Additionally, the background looks unstable. Pedestrians appear to float in and out of thin air, and the red bus oddly changes shape and color.” [sic] when moving across the screen. ”
The video also contains a watermark that matches that of a Facebook page called UK MODE, suggesting it originated from there, although it is not currently visible on that page. However, its Facebook page also shares many similar videos, many of which look convincing at first glance, but also feature AI-generated features such as facial glitches and mismatched backgrounds.
While this particular video may seem clearly AI-like to some, we have seen this clip manipulated by accounts that appear to be taking it at face value.
OpenAI said it investigated the videos and took action against the text-to-video app Sora, including banning the user who created the video.
The company added that it investigates and enforces accounts that engage in harmful deceptive activities, such as misrepresenting the origin of content.
Our guide will help you find AI-generated videos and images. But as AI content becomes more sophisticated and pervasive, we urge people to be careful about the content they see on social media, even if it is very real.
