Sora2 conquers TikTok & Instagram with amazing AI deepfakes – from Jackson to Tupac. But where does entertainment turn into risk? Highlights & Dangers.
Michael Jackson is alive and stealing from KFC guests, Albert Einstein and Tupac enter the ring as MMA fighters, Martin Luther King Jr. runs through parking lots with an unpaid slushy and Kurt Cobain has discovered his love for paragliding. What’s possible with Sora2, OpenAI’s new generative AI app, may be entertaining, but the flood of AI deepfakes also says something about our present — and the future. An overview.
Editorial recommendations
Sora2: The entertaining modeling of reality?
For a few weeks now, fake videos with deceased personalities, talking animals, celebrity conversations, declarations of love and pranks have been circulating on Instagram and TikTok, all of which were created by users with just a few clicks, access permissions and data releases. The reason for this: At the beginning of October, OpenAI, the company behind ChatGPT, introduced its revised video generator Sora2 and the associated app: Sora. The AI generator has already been downloaded thousands of times – and since then the Internet has barely been able to recover from all the content, all of which was artificially generated. The entertainment factor seems to be greater than the dangers behind such deepfakes – or does it?
It is often harmless videos that have the most reach online – such as recordings from police body cameras in which officers control well-known characters such as SpongeBob or Scooby-Doo, or countless variations of Jesus Christ looking for his way in the modern world. Generative reincarnations of deceased audience favorites such as Heath Ledger, Tupac Shakur and Robin Williams are also popular – the possibilities with Sora2 seem endless.
It doesn’t take much to generate deepfakes like these: The model accepts freely chosen text descriptions and uses them to create short video clips in realistic quality.
These are the most spectacular Sora2 videos
- Michael Jackson is alive – and hungry. In this Sora video he steals a KFC guest’s food.
- Here’s Tupac Shakur freestyling in German at the Frankfurt train station.
- OpenAI CEO Sam Altmann steals from Target? At least in this Sora video it does.
- The Internet actually celebrates Bob Ross for his harmonious nature, but this AI-induced, completely different side of the painter is also well received by his fans.
- Stephen Hawking, who warned about the risks of AI technologies during his lifetime, is now doing special half-pipe tricks thanks to them.
All AI — now what?
Despite all the fun, the broad accessibility and simple method of application, coupled with a lack of regulation regarding user data, copyright and personal rights, also raise serious questions. On the one hand, cases have already been reported in which the relatives of deceased celebrities spoke of retraumatization and asked for the videos to be contained.
For example, Zelda Williams, the daughter of the actor Robin Williams, who died in 2014, made it known via Instagram story that these types of videos were anything but entertaining for her as a relative and that her father “wouldn’t think much of it either”.
The descendants of Martin Luther King Jr. also expressed sharp criticism of the spread of deepfakes, in which the civil rights activist’s words were not only twisted, but sometimes converted into racist versions and spread.
Although OpenAI stated in all of these cases that it is making efforts to contain this content, there is a fundamental lack of regulation of such deepfakes on the Internet.
Copyright issues in the entertainment industry
The issue of regulation now plays an important role. As Deadline reported, CEO Charles Rivkin said in a statement: “Since the release of Sora2, videos have appeared in OpenAI’s app and on social media that infringe the rights of our films, shows and characters.” He added that the company must act “immediately and decisively” to stop the copyright infringement.
OpenAI CEO Sam Altman then spoke up. His Declarations According to OpenAI, it has received positive feedback from many rights holders about the “interactive fan fiction”; But he understands why there are concerns about how these characters are used. So the rights holders should now also have the decision as to whether and how their characters can be represented in Sora2. However, exactly how this is implemented is not known.
Generative AI and the concern for…everything?
There are several concerns behind the countless possibilities for creating an alternate reality with Sora2. On the one hand, many fear that the “AI slop” will become even more rapid, i.e. the mass filling of the Internet with AI content that ends up in users’ daily feeds through algorithms controlled by AI. This is already having an impact on the responsible handling of the information provided by these posts: the ability to distinguish real content from fake content is becoming more difficult to grasp.
Plus: Some users specifically exploit the generative video app to spread disinformation, increase the reach of polarizing content and at the same time stoke fear of the power of AI slop. Sora2 also poses a latent threat to political reality. President Trump recently published a statement in response to the “No King” protests in the USA AI generated videowhich shows him with a crown on his head in a fighter jet, firing feces at the demonstrators. Not long before, a deepfake appeared Chuck Schumerthe Senate minority leader, said that the Democrats were just a bunch of “woke pieces of shit.”