A new lawsuit alleges that Apple illegally scraped YouTube videos to train its artificial intelligence models, bypassing the platform’s technical protections. The complaint, filed on behalf of three YouTubers, claims Apple built internal datasets from copyrighted content without consent or compensation.
According to the filing, Apple allegedly “improperly circumvented” YouTube’s safeguards that are designed to block automated harvesting at scale. The creators argue that their videos were ingested into training corpora for Apple’s generative systems, raising questions about copyright infringement, fair use, and compliance with YouTube’s terms of service.
The case targets Apple’s broader data acquisition pipeline for machine learning, focusing on how large language models and multimodal models are fed with online media. It also places Apple in the growing legal battle over dataset provenance that already involves other major AI developers, testing how courts will treat mass scraping from closed platforms.
