Yahoo
1 month agoOpenAI Secretly Trained GPT-4 With More Than a Million Hours of Transcribed YouTube Videos
OpenAI's text-to-video generator, Sora, may have been trained on YouTube videos, raising concerns about data sourcing and copyright infringement.
AI companies like OpenAI are utilizing large amounts of potentially murky data for training models, leading to legal challenges regarding fair compensation and copyright infringement. [ more ]