OpenAI's YouTube Videos Utilisation For Training 'Sora' Violates Rule: YouTube CEO

OpenAI has been frequently accused of utilising third party content for its AI research, sparking this discussion is YouTube Chief Executive Officer Neal Mohan’s statement. 

Mohan expressed concern if the ChatGPT maker is using YouTube videos to train its text-to-video generator 'Sora', suggesting that these actions could potentially breach YouTube's terms and conditions, according to a media report on Friday.

Mohan added although he is not sure if OpenAI is using YouTube’s video to refine Sora's output. But if this is the case he stressed, the AI research company is violating a business policy.

Additionally, he emphasised that OpenAI’s action would also affect the efforts of YouTube creators who share their content on the platform. 

He mentioned that YouTube does not allow the downloading of transcripts or video clips on their platform, and in situations like these, OpenAI's access to user content would be deemed unacceptable if confirmed.

Meanwhile, there has been an ongoing debate surrounding the materials utilised by OpenAI to train the AI models that underpin popular content creation tools such as ChatGPT and DALL-E. 

According to reports, Sora and other generative AI tools operate by assimilating a wide range of content from different online sources for generating new content, including videos, images, storytelling text and more.

Further, Mohan highlighted that Google upholds the individual contracts that YouTube has with creators before determining the use of videos from the platform in training the company's robust AI model, Gemini.