OpenAI Unveils Sora Feed: Personalized AI Video Recommendations with Robust Safety Measures

OpenAI has introduced the Sora feed, a new platform aimed at fostering creativity and exploration within its AI video generation ecosystem. The feed utilizes personalized recommendations to inspire users, emphasizing active participation over passive consumption. Key principles guiding the Sora feed include: * **Optimizing for Creativity:** Prioritizing content that encourages creative expression and active engagement. * **User Control:** Providing steerable ranking, allowing users to tailor the algorithm to their preferences. Parents can also disable feed personalization and continuous scroll for teens via ChatGPT parental controls. * **Prioritizing Connection:** Favoring content that strengthens and forms new connections, especially through interactive “Cameo” flows. * **Balancing Safety and Freedom:** Implementing robust guardrails to prevent unsafe or harmful content while allowing for creative expression. The Sora feed's recommendation algorithms consider various signals to personalize the experience, including user activity on Sora (posts, follows, likes, comments, remixes, and general location), ChatGPT history (with user opt-out options), content engagement metrics (views, likes, comments, remixes), author signals (follower count, past engagement), and safety flags. Safety is a core focus. The platform proactively blocks unsafe or harmful generations, including restrictions on sexual content, graphic violence, extremist propaganda, hate content, and content promoting self-harm. Automated tools and human review are used to monitor content for compliance with OpenAI's Global Usage Policies. Users can also report content that violates these policies. OpenAI acknowledges that balancing safety and freedom is an ongoing process, and user feedback will be crucial in refining the Sora feed's recommendation and safety systems.