OpenAI accused of training Sora on Netflix shows without permission
A Washington Post investigation reportedly found that OpenAI trained its Sora video generation tool on copyrighted Netflix shows, TikTok videos, and studio films without permission. OpenAI claims it used publicly available and licensed data but has not disclosed its training sources. Content creators and copyright holders have raised concerns about unauthorized replication of their proprietary media.
What happened
A Washington Post investigation found that OpenAI trained Sora on copyrighted content without permission, including Netflix shows, TikTok videos, and major studio films.
Sora closely replicates scenes from popular media including Netflix's "Wednesday," video games like "Minecraft," and animated logos from Warner Bros., DreamWorks, and other entertainment studios.
Reported impact
Content creators, copyright holders, and viewers face significant concerns about creative ownership, revenue, and the implications of AI replicating proprietary media without explicit consent.
Index analysis
OpenAI maintains it used a combination of "publicly available and licensed data" to develop Sora but has not disclosed its training sources, citing "competitive reasons."
According to MIT research scientist Joanna Materzynska, who studies AI datasets: "The model reflects the training data. There's no magic involved."
Governance controls
The incident highlights broader challenges in AI development around sourcing data ethically while respecting intellectual property rights.
The incident raises questions for policymakers about regulation of data used to train AI systems.
Classification
Sources and evidence
This record was researched and written by the Index, based on the following original reporting.