OpenAI has launched a new AI-powered video app called Sora that enables users to generate short AI videos, up to 10 seconds long, which can be shared on social media-style streams.
The app uses copyrighted content as a source for generating videos, with a policy requiring copyright owners to opt out if they do not want their content included.
The big picture: This approach extends OpenAI’s precedent from its image-generation tools and has so far caused friction within the entertainment industry.
- Major copyright holders, including Disney, have already opted out to prevent their material from being used in Sora.
- OpenAI has engaged in discussions with various copyright owners to clarify and negotiate its usage policies.
Flashback: Earlier in the year, OpenAI advocated for the Trump administration to recognize training AI on copyrighted data under the “fair use” doctrine, arguing this is vital for U.S. AI competitiveness and national security.
Zoom in: The app incorporates safeguards preventing the unauthorized use of public figures’ likenesses, requiring users’ explicit permission before videos involving their image can be produced or shared.
- A “liveness check” feature prompts users to perform actions like moving their head and speaking random numbers to verify identity for content creation.
- Sora includes a feature called Cameo, which allows users to create realistic AI avatars of themselves and insert these avatars into AI-generated scenes.
