A wave of prominent Australian musicians, including Paul Dempsey of Something For Kate and Bernard Fanning, have expressed outrage after discovering their original songs were included in datasets used to train artificial intelligence models. The discovery came via a dataset search tool recently created by U.S. publication The Atlantic, which revealed millions of creative works have been scraped from the internet to power the technology.
The finding has struck a nerve across the Australian music industry, with Dempsey lamenting that the unauthorized use renders his work "all just rendered useless." It underscores a growing global tension between creators and AI developers, who often treat publicly available works as raw material for training without explicit consent or compensation.
According to The Atlantic's tool, the datasets contain millions of songs scraped from the web, encompassing artists from Nick Cave to Kylie Minogue. The scope of the scrape highlights the sheer volume of copyrighted material being ingested by AI systems, raising serious questions about intellectual property rights in the age of generative AI.
Industry representatives argue that this practice devalues original artistry and could undermine the livelihood of musicians. Without clear legal frameworks or licensing agreements, artists are left with little recourse as their creative output fuels commercial AI products. The Australian music community is now calling for stronger protections and transparency from tech companies.
Critics might contend that AI training on publicly accessible content falls under fair use, especially when the output is transformative. However, for musicians whose life's work is being commodified without consent, the distinction offers little comfort.