Sora 2 Shifts Social Media To AI-Identities Media

OpenAI officially launches the AI video generation model Sora in Suqian, Jiangsu, China, on December 10, 2024. (Photo by Costfoto/NurPhoto via Getty Images)

NurPhoto via Getty Images

On OpenAI’s Sora 2, users broadcast virtual identities with fantastical or deliberately hyper-realistic scenes, which can mean anything but reality.

Compared to existing social media platforms such as Facebook, Instagram, X and Tiktok, Sora 2 is changing what we mean by expressing oneself and connecting with others. When everyone generates, remixes, “cameos” and shares, we won’t know who we are socializing with. We may even get more confused about our true selves, because our likeness can easily be artificialized and consumed by others.

Sora 2 integrates sound, including human speech, directly into the model. This technical development, following Google’s Veo 3, trend of multi-modal AI research from audio-visual segmentation to a sound and sight fusion. As we become the directors and stars of these digital narratives, does this new power veil or expresses ourselves?

Impact Across Industries

The Social Media:

The most disruptive element of Sora 2 is its explicit ambition to become the first AI-identities media platform. Former social media was built on networks of people sharing glimpses of their real lives. AI-identities media is being built on networks of digital personas, where a verified likeness is merely the first asset to be remixed, inserted into any fantasy, and performed across a feed of limitless fictional contexts. Identity itself becomes the modular commodity.

This shift will also recalibrate entire industries.

The Influencer Economy:

The concept of an influencer will be fundamentally unbundled from a human being. We will see the rise of entirely digital personalities, algorithmically optimized for engagement, who never age, never err, and are endlessly malleable.

The Advertising Playbook:

The economics of ad production are being overturned. Why hire a model or crew when you can generate a thousand variations of a scene for a subscription fee? Value will shift from production logistics to creative direction and brand strategy.

AI Entertainment:

Technically, Sora 2 delivers more natural lighting and accurate physics compared to the previous model. It is a medium for conceptual mood pieces and fantastical scenarios, forms where literal human nuance is secondary to visual impact. The model may still struggle with rich human expressions. For instance, the training process requires labeling emotions like “happy” or “smile,” but it cannot grasp its nuances. Is a character’s joy malevolent, nostalgic, or triumphant? AI video models can replicate the symptom of an emotion, but not the context that gives it meaning.

Limitations and Risks

AI videos will catalyze a new genre of entertainment built on surreal visuals and rapidly iterated concepts. This may accelerate the industry’s pivot towards shorter formats, catering to dwindling attention spans. Yet, herein lies the new frontier for the artist. In an age of mass-produced, consumable visuals, the most powerful artistic statement may be to go against the stream, to use these very tools to create works that demand thinking and reflection rather than fast food type of media consumption.

A society saturated with synthetic narratives risks losing its common ground. When we scroll through endless feeds of digital avatars in impossible scenarios, we lose touch with the shared human experience. This technological isolation could exacerbate polarization, as we retreat into fictional worlds rather than engaging with truths.

The greater risk is the “liar’s dividend”, the growing skepticism that makes us doubt everything we see. A genuine plea or a documented event could be dismissed as just another clever fake, eroding the last vestiges of our shared reality.

One proposed safeguard is robust watermarking. While not a panacea, clear labels for AI-generated content are a necessary first step to maintain a baseline of trust. They are the digital equivalent of a disclaimer, allowing consumers to contextualize what they are watching.

AI videos are mediating what we see each other and how we see ourselves. But in giving us the power to fabricate reality so effortlessly, it challenges us to remember what is real. The future of our digital society w ill depend not on our ability to generate perfect videos, but on our courage to preserve imperfect, genuine human connection amidst spectacles.