In a recent interview, Mira Murati stated, “Some creative tasks that require creativity may disappear due to productive artificial intelligence, but perhaps they should never have emerged in the first place.” Murati believes that some creative roles are inherently inefficient or unnecessary. She fundamentally views the elimination of certain creative tasks by artificial intelligence as a positive development and says, “I believe using productive AI tools for education and creativity will expand our intelligence.”

Murati had previously sparked controversy with a evasive response to a question focused on Sora. When asked in an interview whether Sora had been trained on YouTube videos, Murati hesitated for a moment before surprisingly stating, “I’m not entirely sure.” Murati clarified that they used publicly available videos and privately licensed video sets for training the system, neither directly acknowledging nor outright denying the high likelihood of YouTube videos being used in Sora‘s training.

Given the impressive capabilities of the current system, it seems highly improbable that it was developed without leveraging support from YouTube videos. OpenAI generally considers publicly available content as freely usable for training large language models (LLMs). However, YouTube CEO Neal Mohan expressed uncertainty about whether OpenAI used YouTube videos to train Sora, noting that such usage would clearly violate the platform’s policies. In the near future, YouTube could potentially pursue a serious legal case against OpenAI regarding this matter.