YouTube creators surprised to find Apple and others trained AI on their videos

YouTube creators were surprised to discover that major tech companies like Apple, Salesforce, and Anthropic had trained their AI models on tens of thousands of YouTube videos without the creators' consent. The companies utilized a dataset called "the Pile," created by EleutherAI, which includes YouTube captions scraped from over 48,000 channels, including videos from popular YouTubers like MrBeast and PewDiePie. The dataset also contained content from mainstream media brands like Ars Technica. While some creators expressed frustration at the unauthorized use of their content, companies like Anthropic defended their actions, stating that the dataset used was a small subset of YouTube subtitles and did not directly violate YouTube's terms of service. This incident sheds light on the challenges creators face in controlling how their content is used online and raises questions about the ethical implications of training AI models on publicly available data.