A question posed by OpenAI CTO Mira Murati in recent months was, “Was Sora trained with YouTube videos?” There was an evasive answer to his question.
OpenAI CTO Mira Murati asked by Joanna Stern last month, “Was Sora trained with YouTube videos?” After thinking about the question for a while, surprisingly, “I’m not exactly sure about that.” your answer had given. Murati stated that they used publicly available videos and privately licensed video sets in training the system. He does not accept it directly in order to avoid legal problems.Interestingly, he does not deny it, but it is very likely that Sora’s training was aided by millions of videos on YouTube (Google may not like this at all and may file an official lawsuit if it finds evidence in the coming period.). It does not seem possible for the system to reach its breathtaking current state without support from videos on YouTube.OpenAI generally uses publicly available content LLM (Large Language Model) thinks that it can be used free of charge in education. This issue came up once again today because OpenAI COO Brad Lightcap directed towards, “Was Sora trained with YouTube videos?” He avoided answering the question.
Speaking about this, YouTube CEO Neal Mohandoesn’t know if OpenAI used YouTube videos to train Sora, but He states that such a situation would be a clear violation of the platform’s policies. YouTube will file a serious lawsuit against OpenAI on this issue in the coming period. can open. If you missed it, the company’s new artificial intelligence model “Sora”, It can produce 60-second videos from text And It produces very, very good results. The company, which entered this field a little late but has prepared a very ambitious infrastructure, states that Sora will be opened to certain people at the first stage against possible security risks. They said that they would make the public exit open to everyone later in order to prevent the abuse of the system and According to the statement made by Mira Murati, it aims to start public use in the coming months. The company states that Sora can produce complex scenes with a large number of characters / people / creatures / objects, and the first examples shown confirm this.