OpenAI Is Set to Introduce a Social Platform for AI-Created Videos

OpenAI is gearing up to roll out a standalone app for its video generation AI model Sora 2, as reported by WIRED. The app features a vertical video feed with swipe-to-scroll functionality, resembling TikTok but with entirely AI-generated content. A For You–style page driven by a recommendation algorithm enhances the experience. A menu bar on the right allows users to like, comment, or remix videos.

According to documents seen by WIRED, users can create video clips lasting up to 10 seconds using OpenAI’s advanced video model. However, there is no capability to upload images or videos from a user’s camera roll or any other applications.

The Sora 2 App includes an identity verification feature, enabling users to confirm their likeness. If verified, users can have their likeness used in videos. Other users can tag them and incorporate their likeness into clips. For instance, someone could produce a video of themselves enjoying a roller coaster ride with a friend. Users will receive notifications whenever their likeness is utilized—even if the clip is never officially posted and stays in draft form, sources indicate.

OpenAI launched the app internally last week, garnering overwhelmingly positive feedback from employees, as per documents obtained by WIRED. Employees have engaged with the tool so much that some managers have humorously noted it could affect productivity.

OpenAI opted not to provide comments.

It seems OpenAI is betting that the Sora 2 app will transform how people engage with AI-generated video—much like ChatGPT did for AI-generated text. Internally, there’s also a sentiment that President Trump’s fluctuating deal to sell TikTok’s U.S. operations has presented OpenAI with a unique chance to launch a short-form video app, especially one with no strong links to China.

OpenAI officially introduced Sora in December of last year. Initially, it was accessible solely through a webpage but was soon integrated directly into the ChatGPT app. At that time, the model was among the most advanced AI video generators, although OpenAI acknowledged certain limitations, such as a lack of understanding of physics and challenges in creating realistic action scenes, particularly in longer videos.

The Sora 2 app will face competition from emerging AI video solutions from major tech players like Meta and Google. Just last week, Meta unveiled a new section in its Meta AI app called Vibes, designated entirely for creating and sharing short AI-generated videos. Earlier this month, Google revealed it would incorporate a bespoke version of its latest video generation model, Veo 3, into YouTube.

Conversely, TikTok has adopted a more cautious stance towards AI-generated content. The platform has recently redefined its guidelines regarding the type of AI-generated videos it permits, now explicitly prohibiting content that is “misleading about matters of public significance or harmful to individuals.”

Sources indicate that the Sora 2 app often declines to generate videos due to copyright protections and various filters. OpenAI is currently contesting several lawsuits regarding alleged copyright infringements, including a notable case initiated by The New York Times, which claims that OpenAI trained its models using the paper’s copyrighted material.

Additionally, OpenAI faces increasing scrutiny concerning child safety issues. On Monday, the company launched new parental controls, allowing parents and teenagers to link their accounts. They also stated they are developing an age-prediction tool that would automatically direct users believed to be under 18 to a more restricted version of ChatGPT, which limits certain interactions. It remains unclear what age restrictions may be implemented in the Sora 2 app.

This is an edition of the Model Behavior newsletter. Read previous newsletters here.