A new tool from The Atlantic, titled AI Watchdog, is once again drawing attention to the growing controversy surrounding the use of artists’ work in AI training datasets as generative technology continues to expand across the music industry. The controversy follows the launch of publicly searchable platforms that allow musicians to determine whether their work appears in datasets linked to AI training.
Research compiled by journalist Alex Reisner highlighted four major datasets containing more than 21 million songs from both internationally recognised artists and independent creators. While the databases themselves do not identify which companies may have used the material, several AI developers and research groups have previously acknowledged relying on comparable datasets to train their systems.
Artist frustration has quickly moved beyond abstract concerns about copyright. Singer SZA recently revealed that a search of the datasets appeared to show 238 of her songs included in AI training material, raising concerns that even unreleased work may have been swept into the process. Producer Kenneth Blume also condemned AI music companies, criticising what he described as a system that threatens the livelihoods of working musicians.
Major record labels and music organisations have already pursued legal action against leading AI music platforms over allegations of copyright infringement and unauthorised use of recordings. Meanwhile, the wider industry remains split between those embracing AI as a creative tool and those who see current practices as a direct challenge to artistic ownership.
