Apple and Nvidia used YouTube videos without the authors' permission to train artificial intelligence

Apple and other tech giants such as Nvidia and Salesforce have used subtitles from YouTube videos to train their artificial intelligence models without permission from the content creators.

Here's What We Know

According to a new report, the subtitles from the video were compiled into a file by EleutherAI, a firm that helps developers train AI models. The original goal, according to the report, was to create training materials for small developers and scientists. EleutherAI collected captions from more than 170,000 videos, including contributions from well-known technobloggers and comedians.

However, large companies such as Apple have used this dataset created by EleutherAI, called YouTube Subtitles, which does not include images but contains simple text subtitles for videos. This set also includes translations in languages such as Japanese, German, and Arabic.

Among the videos used were some that have already been removed from the platform. Some authors who found their works in these models expressed dissatisfaction as they were not asked for permission. Companies such as OpenAI and Meta have defended their actions by citing the fair use doctrine, which allows copyrighted material to be used in certain situations without a licence.

Lawyers and public figures have expressed concern that tech companies are acting without the consent of authors, raising questions about the rights and ethics of content use.

Source: Wired