The video-sharing platform YouTube warned the company OpenAI not to use videos from their platform to train their AI models.
YouTube CEO Neal Mohan said in an interview with the American newspaper Bloomberg that using platform-specific videos to train OpenAI models would definitely break the rules.
Mohan added: “From the content creator’s perspective, when they bring their serious work to our platform, they have certain expectations, including that the terms of service will be complied with, which do not allow downloading videos or accompanying texts, as that would be a clear violation of our terms of service. Those are the rules of the road regarding content on our platform.”
OpenAI, active in the field of artificial intelligence, recently announced the Sora model, a model that converts text commands into short videos based on artificial intelligence.
The unveiling of the Sora model has sparked mixed reactions amid OpenAI’s ambiguity and lack of disclosure of the data sources used to train the model.
Last March, the company’s technology director, Mira Murati, contributed to the criticisms directed at the company by stating to The Wall Street Journal that she was not sure if the Sora model relied on data from YouTube, Instagram, or Facebook.
The American newspaper Wall Street Journal recently reported that OpenAI plans to use the accompanying texts of YouTube videos to train its upcoming model GPT-5.
On the other hand, Google, which competes with OpenAI in the field of artificial intelligence, claims to respect the terms of use when it comes to the YouTube platform it owns. Google’s artificial intelligence model Gemini requires similar data for training and learning.
Mohannad claimed that Google only uses specific video clips based on the permissions provided in each content creator’s licensing agreement.