News reports allege, OpenAI has trained its artificial intelligence model through video content collected from YouTube, but has not been stopped by Google. The reason seems to be related to that Google itself also trains its artificial intelligence model through video content from YouTube.
Prior to this, YouTube CEO Neal Mohan was interviewed by Bloomberg News and believed that OpenAI trained itsSora, a tool for generating videos from textThe artificial intelligence model behind it clearly violates YouTube's usage policies.
Google spokesman Matt Bryant explained that unauthorized downloading of YouTube videos violates Google's usage policies, but it has not yet been confirmed whether OpenAI has similarly violated its usage policies.
However, a New York Times report pointed out that Google also uses YouTube video content to train its artificial intelligence models. At the same time, Google explained that it currently only uses video content created by users who agree to participate in the experimental program for training, and does not use other video content that has not joined the experimental program to train artificial intelligence models.
Prior to this, Google had adjusted in July 2023Service Privacy PolicyThis will make it easier for Google to collect user-uploaded public information for use in AI model training. Compared to the original description, which emphasized the use of public information for natural language model or application service training, Google's new adjustment clearly aims to clarify that all future user-uploaded public information can be "reasonably" used for AI project training.
