What if the future of social media no longer depended on human creators, but on artificial intelligence capable of endlessly generating videos, trends, and even virtual influencers? That’s the rumor circulating about OpenAI: the company, already at the heart of the revolution in language and image models, is reportedly working on a generative video platform designed as an alternative to TikTok.
It’s a bold idea. While TikTok relies on user creativity and the power of its recommendation algorithm, OpenAI could take things a step further by generating content itself using its multimodal models. Imagine an endless stream of short videos—scripted, edited, and performed by artificial avatars—capable of mimicking the viral trends of digital culture.
This hypothesis raises some staggering questions: What would be the implications for the attention economy, for content creators, and for social media regulation? Would such a platform mark the birth of an artificial social network, where humans would be nothing more than spectators?
A completely fabricated narrative
This project would build on recent advances in multimodal models: text, images, audio, and video can now be generated by a single system. In practical terms, a user could ask, “Show me a funny video of a character dancing on a volcano,” and within seconds, a perfectly coherent sequence featuring an artificial avatar would appear.
This stream would no longer depend on human-generated content, but on an infinite reservoir of content generated on demand—or even anticipated by AI based on the user’s implicit preferences. Personalization would become virtually total, with each viewer receiving a unique, tailor-made stream.
The opportunity is enormous: TikTok claims it will have over 1.7 billion monthly active users by 2025, who spend an average of 95 minutes a day on the platform1. Replicating this model by in-sourcing production through AI would allow OpenAI to position itself in a colossal market dominated by the attention economy.
The end of the central role of human creators?
The key difference in such a project lies in the role given to humans. TikTok and Instagram Reels thrive thanks to creators who produce content, experiment, and engage with their communities. In an artificial TikTok, these creators would be replaced by virtual avatars programmed to charm, move, or entertain.
This idea isn't so far-fetched. The virtual influencer market already exists: according to Allied Market Research, it was worth $21 billion in 2024 and is expected to grow by more than 35% annually through 20302. Artificial characters like Lil Miquela or Imma, followed by millions of people, show that the public is ready to form attachments to fictional identities.
If OpenAI launches such a platform, it could create hundreds, or even thousands, of “artificial stars,” each optimized to capture a specific audience segment. This could marginalize the role of human creators or force them to collaborate with AI in order to survive in this ecosystem.
An economic opportunity… and a risk to attention
The short-form video market is a major player. According to Fortune Business Insights, it could reach $475 billion by 20303, largely driven by targeted advertising and sponsored content. By controlling both the creation and distribution of videos, OpenAI could capture not only attention but also the entire advertising value chain.
But this prospect is cause for concern. An unlimited stream of AI-generated videos, constantly optimized to maximize screen time, could amplify the criticisms already leveled at social media: addiction, loss of concentration, and cognitive overload. In a world where videos are optimized by AI, human attention would become, more than ever, a resource exploited on a massive scale.
The shadow of misinformation and falsehoods
Such a project would not be without societal risks. While AI can generate lighthearted entertainment, it can also create misleading content. The proliferation of videos indistinguishable from reality would open the door to massive disinformation campaigns. Deepfakes, which are already problematic, would become commonplace in an environment where every video is artificial.
Technological safeguards, such as digital watermarks or traceability systems, would be essential. But their effectiveness remains limited: as early as 2023, a study by the University of Cambridge showed that 72% of users had difficulty distinguishing a deepfake from an authentic video4.
Transparency then becomes a major issue: how can we clearly inform users that what they are viewing was not created by a human? And, more importantly, will this actually change their relationship with the content?
Toward an “artificial flow society”?
This project illustrates a broader trend: the rise of what some researchers call the “artificial flow society.” In this model, AI no longer merely recommends human-created content; it becomes its primary producer. Digital culture is thus fueled by algorithms rather than collective creativity.
This raises a philosophical question: do we prefer content that is true, limited, and sometimes imperfect, or content that is artificial, unlimited, and always tailored? Authenticity remains a core value, but the power of AI could redefine expectations: no longer “true,” but “engaging.”
This shift raises questions about the very role of human creativity in the attention economy. In the future, will our video feeds still reflect our societies, or will they be the expression of algorithms shaping our collective imaginations?
References
1. Statista. (2025). TikTok user statistics worldwide.
https://www.statista.com/statistics/1267892/tiktok-global-users/
2. Allied Market Research. (2024). Virtual Influencer Market Size and Forecast.
https://www.alliedmarketresearch.com/virtual-influencer-market
3. Fortune Business Insights. (2024). Short Video Market Size, Share & Forecast.
https://www.fortunebusinessinsights.com/short-video-market-10452
4. University of Cambridge. (2023). Deepfake detection challenges.
https://www.cam.ac.uk/research/news/deepfake-detection-study
