Video generation welcomes the "ChatGPT moment"! OpenAI launches social application to directly compete with TikTok and Meta

According to Zhitong Finance APP, artificial intelligence company OpenAI officially launched a brand new independent social application "Sora" on Tuesday, allowing users to generate and share AI videos and interact with friends. The Sora app is currently invite-only and was first released on the Apple iOS platform, with plans to expand to the Android system in the future. The app is based on the upgraded Sora 2 video generation model, enabling users to generate short videos through text prompts and browse content created by others. The app also introduces a "virtual avatar" feature, allowing users to create highly realistic AI avatars and voices, which can be inserted into videos made by friends with permission. Despite ChatGPT attracting over 700 million users weekly, OpenAI still faces fierce competition in the video generation field. Companies like Google (GOOG.US, GOOGL.US), Runway AI, and Midjourney also offer AI video tools, some of which generate content faster and at lower costs. The release of Sora marks an important step for OpenAI in creating social media products, directly competing with TikTok and Meta (META.US), which recently launched the AI video stream "Vibes." Analysts believe Sora could open new channels for advertising revenue for OpenAI and further enhance its technological reputation. OpenAI engineer Thomas Dimson acknowledged in a media briefing that there was initial skepticism about AI-generated video streams, but the "virtual avatar" feature convinced the team that this could have a positive impact on the world. He pointed out, "Many social platforms have gradually deviated from connecting people, and we hope Sora can bring back that interaction." Technically, Sora 2 aims to solve two long-standing challenges in AI video generation: physical laws and scene transitions. OpenAI stated that the new software can more accurately present effects like fluid dynamics and buoyancy, and can more faithfully follow user prompts in multi-camera videos. Additionally, Sora 2 can automatically stitch scenes together and use AI to generate multilingual dialogues, sound effects, and background noise. Bill Peebles, head of the Sora team, said, "We believe this could be the ChatGPT moment for video generation." To address potential misuse, OpenAI emphasized that Sora will not allow the generation of videos involving public figures, and all videos will carry watermarks indicating they are AI-generated. At the same time, the app has blocked screen recording features to limit external sharing of videos