OpenAI Launches AI Video App That Edits Copyrighted Content Automatically
GeokHub
Contributing Writer
OpenAI has launched Sora, a new AI video app that lets users create short AI-generated videos — including ones spun from existing copyrighted material — and share them to social-feed style streams.
Sora produces clips up to 10 seconds long and includes a feature called Cameo that allows people to generate realistic AI versions of themselves and insert those likenesses into generated scenes. To protect against misuse of personal likenesses, the app requires a “liveness check” before allowing a person’s image to be created: users must move their head on camera and recite a random string of numbers so the system can verify presence. OpenAI also restricts creation of videos of public figures or other people unless those individuals first upload an AI-generated video of themselves and give permission.
OpenAI says its approach to copyrighted material follows the company’s prior opt-out policy for image generation: copyright holders must opt out if they do not want their works to appear in Sora’s video feed. The company has been in talks with rights holders about the policy and some studios have already chosen to opt out. OpenAI is also exploring attribution and revenue-sharing mechanisms as part of its efforts to address rights holders’ concerns.
The release of Sora raises legal and industry questions about how copyright law applies to AI remixing and whether additional licensing or guardrails are required. OpenAI previously sought clarification from U.S. authorities on whether training AI models on copyrighted material falls under fair-use protections — an argument it has framed as important to U.S. competitiveness in AI.
Industry observers say Sora could pose strong competition to established social and video platforms by combining short-form video, AI editing tools, and seamless sharing. At the same time, legal scrutiny from media companies and regulators seems likely as creators, studios, and lawmakers weigh the implications for licensing, attribution and misuse.
