Overview

• Apple denies using transcribed YouTube videos for its Apple Intelligence AI, claiming it uses licensed data.
• The dataset, compiled by EleutherAI, includes subtitles from over 173,000 videos across 48,000 channels.
• Affected creators include prominent YouTubers like MrBeast, Marques Brownlee, and educational channels like Khan Academy.
• The unauthorized use of YouTube data violates the platform's terms of service, raising questions about ethical AI practices.
• Tech companies argue they are not at fault as they did not directly scrape the data themselves.