PEOPLE have been left "terrified" by a series of seemingly ordinary videos.
But can you tell what's wrong with them in 30 seconds?
"Im watching every video on that website in disbelief," one X user wrote after discovering the truth behind them.
"This is wild," another person wrote.
Well, the truth is none of these clips are real - they were all made by AI.
Artificial intelligence reaches major milestone 'for the first time ever'
OpenAI, the company that shocked with the world with ChatGPT, has unleashed a more advanced version that can now generate videos.
Much like previous models that generate images, all a person has to do is describe in writing what they want and the AI will instantly create it.
The tech - known as Sora, the Japanese word for “sky” - can produce realistic footage up to a minute long.
The idea that an AI can create a hyper-realistic video of, say, a politician doing something untoward should ring alarm bells as we enter into the most election-heavy year in human history
Dr Andrew Rogoyski
It is also capable of creating a video based on an image or extend existing footage with new material.
But much like the AI generated images we've seen in recent years with extra fingers and teeth, some of the videos need some work.
Take the example from OpenAI below of a cat waking its owner which has two strange errors - can you spot them?
Firstly, the woman's arm disappears when she rolls over.
And secondly, the cat grows an extra arm and paw on its left side.
Dangers
The service isn't available to the general public just yet.
Inside home of the future - including AI baby crib
For now, only selected researchers and video creators are being invited to test it for any dangerous mistakes, such as misinformation, hateful content, and bias.
Dr Andrew Rogoyski from the University of Surrey said: "OpenAI has recognised the potential for harm with such a system.
"The idea that an AI can create a hyper-realistic video of, say, a politician doing something untoward should ring alarm bells as we enter into the most election-heavy year in human history, with over 60 democratic elections in '24 and half the planet's population voting.
"Interestingly, OpenAI plans to watermark Sora's outputs with C2PA, a digital certification system that is growing in popularity as a means to track the provenance of information."
In a web post, OpenAI explained: "We’ll be engaging policymakers, educators and artists around the world to understand their concerns and to identify positive use cases for this new technology.
"Despite extensive research and testing, we cannot predict all of the beneficial ways people will use our technology, nor all the ways people will abuse it."
