OpenAI’s Sora 2: AI-Driven Social Video Platform with Privacy Concerns

OpenAI is moving toward a new social ecosystem where chat tools are turning into full-fledged video platforms. Recent rumors reveal the concept of Sora 2 – a standalone app with a TikTok-style feed, where all videos are created by artificial intelligence.

Within this service there is a familiar navigation mechanism: vertical viewing by swiping up, a personalized recommendation algorithm, and interactivity – the ability to like, leave comments, and remix posts. The main difference is that the entire visual content is generated by AI: the model can process text, photos, or existing video and transform them into new video clips lasting up to about 10 seconds, which is tied to the current generation’s technical limitations.

Note that uploading photos or videos without prior editing inside Sora 2 is not provided – all materials are subject to internal processing by the platform.

A Special Social Aspect and Privacy

Although the concept sounds innovative, implementing the social component raises quite a few privacy questions. According to the plan, the app may request identity confirmation using facial recognition to verify similarity. After confirmation, the likeness may be used in a video, and other users will be able to tag the person and apply their appearance in their own materials. Most such uses will be tracked each time the likeness is used, even if the generated video is saved as a draft. Whether this becomes public practice is a question that requires careful decisions by developers and regulators.

The ability to introduce opt-out options or restrictions on the use of another person’s likeness is also discussed, but privacy and data protection issues remain open and require clear norms.

Content Safety and Regulatory Restrictions

Within the framework of the concept, mechanisms for content protection are expected: AI should refuse to create materials that infringe copyrights, and filters will also be applied to restrict certain categories of videos. However, system testing indicatively shows that the risk of prohibited content does not disappear entirely even at lower probabilities.

Media reports indicate inquiries to OpenAI regarding plans for Sora 2, but as of publication no response had been received. There are rumors that the launch of Sora 2 could coincide with the GPT-5 announcement. The project remains in a conceptual stage, but the idea of a fully AI-driven social feed underscores the value of isolating AI content within special frameworks.

Regardless of whether Sora 2 becomes a reality, the question of balancing innovation, privacy, and content rights remains. Clearly, OpenAI will seek to maintain control over the use of likenesses and implement regulatory and technical restrictions to minimize risks for users and creators.

According to Wired, the app will ask users to verify their identity using facial recognition to confirm likeness. After confirmation, their likeness may be used in a video, and other users will be able to tag them and apply their appearance in their materials.

All of these aspects point to the need for careful regulatory and technical handling of future AI-powered social content: its benefits in education and entertainment should be balanced with effective protections for privacy, copyrights, and user safety.