We have previously told you that OpenAI uses YouTube videos to train GPT-4. we had transferred. However, a new claim has been put forward on this issue.
Apple, Anthropic big technology companies like YouTube It was claimed that they trained artificial intelligence systems using subtitles taken from their videos. However, when it was revealed that this data set was collected without YouTube’s permission, allegations that it violated the platform’s terms of service were also brought to the fore.
Research conducted jointly by Proof News and Wired according to, these companies have widely used captions from YouTube videos to develop their AI systems. The research revealed that more than 170 thousand videos were included in this data set and that these contents belonged to more than 48 thousand different YouTube channels. It was emphasized that the data set only included subtitles of the videos.
It was revealed that especially the videos of popular content creators MrBeast and Marques Brownlee, as well as the content of major news organizations such as ABC News, BBC and The New York Times, were also included in this data set. However, at this point, it should be underlined that such data collection and use is against YouTube’s terms of service.
YouTube CEO Neal Mohan had previously emphasized that using video subtitles without permission is unacceptable and that this situation is against the platform’s security and user privacy principles.
As part of the research Proof News It also released an interactive tool. This tool allows users to mine this dataset for their own content or their favorite YouTubers’ content.
Source link: https://webrazzi.com/2024/07/17/teknoloji-devlerinin-youtube-videolarini-izinsiz-kullanarak-yapay-zekayi-egittigi-ortaya-cikti/