Three YouTube creators have filed a proposed class-action lawsuit accusing Apple of improperly scraping their videos to help train AI models. The complaint, filed in federal court in California, alleges Apple used or benefited from datasets built from YouTube videos without creators’ permission.

According to reports, the plaintiffs include the operators behind channels such as Ted Entertainment, Matt Fisher and Golfholics. The lawsuit claims Apple “improperly circumvented” YouTube protections in order to collect or make use of creator-uploaded content for model training, and argues that the company violated copyright and anti-circumvention rules in the process.

Lawsuit centers on training data and platform protections

At the center of the case is the question of how the videos or data derived from them were obtained and whether that process violated YouTube’s terms or U.S. copyright law. The filing reportedly argues that Apple either directly or indirectly relied on large scraped datasets containing YouTube material to train AI systems.

This dispute also revives concerns first raised in 2024, when reporting found that subtitle and transcript data from YouTube videos had appeared in widely circulated AI training datasets used by multiple companies. At the time, Apple said one of the cited models, OpenELM, was a research model and not part of Apple Intelligence.

Apple’s AI data practices remain under scrutiny

Apple has publicly said its current foundation models are trained using a mix of licensed material, synthetic data and publicly available web data gathered through “responsible web crawling.” In its 2025 technical documentation, the company said Apple Intelligence models were built from large-scale multilingual and multimodal datasets, though it did not detail every underlying source.

That gap between broad disclosures and creator-specific consent is becoming a recurring issue across the AI industry. As more lawsuits target the use of online media for model training, courts are increasingly being asked to decide whether public availability is the same as legal permission.