“Until the major labels go through their lawsuits, there’s no way for artists or labels to fight back”: A massive music dataset is allegedly allowing AI to train on 12 million+ songs without permission

AI Watchdog reporting indicates AI models are being trained on massive music datasets (12M+ tracks) without permission, with downloads by major players and datasets sourced from places like Free Music Archive. Ongoing lawsuits target Suno/Udio; APRA AMCOS is launching an investigation, and many indie artists say their catalogs were scraped. This has direct implications for music rights, licensing, and AI-generated content workflows that affect game audio creators and studios.

Read More: https://musictech.com/news/industry/ai-12-million-dataset-training/