A man is flying on a kite, the wind makes both his shirt and hair flutter. While he is flying over a stormy sea, he tells us that a new service that "gives you the power to step into any world or scene" is being introduced.
In another clip in the advertisement video released on Tuesday, Open AI's CEO Sam Altman, seemingly creating a miniature orchestra of clay figures, says that it's not just about productivity:
It's about creating new opportunities.
"Extremely realistic"
Altman doesn't say those words for real. The video – and the sound – is entirely created by the video generator Sora 2, which will soon be made available to the public.
Yes, it's extremely realistic, says Andreas Jahrehorn Önnerfors, professor of history of ideas at Linnaeus University and active at the Media Institute Fojo with a focus on influence, fact-checking and disinformation.
For the ordinary media consumer, it will not be possible to distinguish whether it is human or synthetically produced, he continues.
Used politically
Artificial intelligence is already being used to construct disinformation with political purposes.
US President Donald Trump recently published an apparently manipulated clip, where it looks like Democrat Chuck Schumer says that the party wants more illegal immigrants in the country to get more votes.
According to British media , AI-generated content portraying Muslims as threatening has been spread ahead of protests against immigration.
After tens of thousands of Britons gathered for a right-wing extremist manifestation in London earlier in September, images of similar protest marches against migration in other European cities have been spread – without them having taken place.
Imagination sets limits
Andreas Jahrehorn Önnerfors sees a risk that the progress of technology will impoverish democracy.
We build our decisions on the information that is available. If it becomes more and more manipulated, then we base our decisions on something that is not anchored in reality.
He chooses, however, a cautious optimism for the future. Propaganda is nothing new. Certainly, Sora 2 provides increased accessibility and better opportunities to manipulate content, but it will probably be used in a fairly predictable way.
In the end, the use of these powerful AI generations is based on someone coming up with the crazy idea of having a symphony orchestra of clay figures.
The first Sora model was launched in February 2024. On Tuesday, the sequel was announced.
Sora 2 is so far only available in North America and only for those with special invitations. The material it can produce is described as more realistic.
Rightsholders can actively choose not to have their material used to train the generator, which generates video material from text instructions. The lengths of the film clips are currently limited.
Pornography and extreme violence will be restricted.
Source: Open AI
