OpenAI’s Sora 2 incorporates extensive safety measures focusing on provenance, likeness protection, teen safety, and harmful-content filtering. In a March 23, 2026 update, the company emphasized that every video generated includes provenance signals and C2PA metadata, along with visible moving watermarks to indicate creators. Users must confirm consent to upload images of individuals, with enhanced moderation for minors and shared content.
Additionally, Sora 2 offers users controls over their characters, allowing for management of likeness use and restrictions on major alterations. Protections for young users include filtered feeds and limited direct messaging from adult accounts, alongside parental control features in ChatGPT. OpenAI also implements robust harmful-content checks during both creation and distribution phases, monitoring video and audio outputs to eliminate inappropriate material. Users retain rights to manage shared content and report abuses. The update outlines built-in controls without providing external effectiveness evaluations, emphasizing product safety and user empowerment.
Source link
