Tech giants like Apple have reportedly trained AI models using subtitle files from YouTube videos without the creators' consent. This was done by a third party downloading subtitle files from over 170,000 videos, impacting creators like Marquees Brownlee (MKBHD) and PewDiePie. Despite YouTube's rules against harvesting materials without permission, companies like Apple, Nvidia, and Salesforce used these subtitle files to train AI models. While the aim may have been to provide training materials to developers and academics, the dataset was also utilized by Silicon Valley heavyweights. This raises concerns about the legality of using publicly-available datasets to train AI systems, highlighting the risks of using material without permission. Apple has not responded to requests for comment, but it is clear that the use of unauthorized data in AI training poses ethical and legal challenges.