It is not known for certain whether the new OpenAI neural network has used YouTube videos, but the video platform’s CEO assures that this would be a clear violation of the terms of use.
«If a creator uploads their work to our platform, they have certain expectations», — said Neil Mohan in interview to Bloomberg Originals host Emily Chang. «And one of those is — compliance with the terms of service. Uploading video snippets or transcripts would be a clear violation of».
Sora — is new neural network OpenAI capable of generating realistic short videos based on text queries. Like the startup’s previous products — ChatGPT and DALL-E —, the technology has faced criticism for using various data from the Internet, including potentially copyrighted data, for training.
Last month, OpenAI Chief Technology Officer Mira Murati told the Wall Street Journal in an interview that not sure if Sora learned from user-generated YouTube videos, Facebook, and Instagram. This week, the publication also reported that OpenAI has been discussing training for its next-generation large-scale language model, GPT-5, on transcriptions of publicly available YouTube videos.
Mohan says Google will honor YouTube’s individual contracts with creators when deciding whether to use videos from the platform to train its powerful Gemini AI model.
«Creators have different types of license contracts for content on our platform,» Mohan said.
He added that while «some part of YouTube» may be used to train models such as Gemini, Google and YouTube ensure that this «complies with any terms of service or contract that the» authors have signed.
OpenAI, meanwhile, is running negotiations with major Hollywood studios, agencies and producers about the possibilities of using Sora. According to The Wall Street Journal, the neural network will become available to the public in 2024, and later will be able to support audio and video editing, to make the scenes even more realistic.