OpenAI is developing a standalone social media application powered by its upcoming Sora 2 video model that will feature exclusively artificial intelligence (AI)-generated content.
The app will reportedly feature a vertical video feed with swipe-to-scroll navigation like TikTok’s interface. Unlike traditional social platforms, users will not be able to upload photos or videos from their devices’ camera rolls, according to a report in Wired.
Sora 2 will be restricted to generating clips of 10 seconds or less within the app, though limits for the model outside the application remain unclear. By comparison, TikTok currently allows videos up to 10 minutes in length, a significant increase from its original 15-second limit.
The platform will include an identity verification feature that enables Sora 2 to generate videos using a user’s likeness. Other users could then tag and incorporate verified users’ likenesses when remixing videos. As a safeguard, OpenAI will notify users whenever their likeness is used by another person, even if the resulting video is never posted publicly.
Wired reports the software will block certain video generation requests because of copyright restrictions, though the extent of these protections remains uncertain. The Wall Street Journal has reported OpenAI will require rights holders to actively opt out of having their content appear in Sora 2-generated videos.
The timing of the app’s development may be tied to the ongoing uncertainty surrounding TikTok’s U.S. operations. Oracle Corp., private equity firm Silver Lake, and Abu Dhabi sovereign wealth fund MGX will control roughly 45% of TikTok’s U.S. business under a deal expected to be finalized soon.
Under the deal, TikTok’s Beijing-based parent company ByteDance will retain a 19.9% stake. The remaining 35% will be distributed among new and existing investors, CNBC reported, citing sources familiar with the arrangement.
The social component could also serve as a strategic moat for OpenAI, potentially discouraging users from switching to competing AI video models by tying them to the platform’s emerging community.
