Music: a label to distinguish tracks generated by artificial intelligence
The global music industry is launching two labels: ‘AI-generated’ and ‘AI-assisted’
The music industry is trying to label artificial intelligence. And to do so before the proliferation of synthetic tracks makes it increasingly difficult to distinguish between a recording made by artists and a product generated by an algorithm.
IFPI, RIAA, A2IM, WIN, IMPALA, the Grammys, SAG-AFTRA and the Human Artistry Campaign have introduced a voluntary classification system for sound recordings based on two categories: ‘AI-generated’ and ‘AI-assisted’. The aim is to achieve global adoption by digital music platforms, distributors and other stakeholders in the industry.
Not just a simple label, then, but an attempt to establish a common industry standard in a market where the supply of AI-generated music is growing at a rate that is hard to ignore. In April, Deezer reported that AI-generated tracks accounted for 44 per cent of all new music uploaded to the platform. Apple Music, according to the associations, reportedly found that more than a third of uploaded tracks could be classified as ‘100 per cent AI’.
The first label will be applied when generative artificial intelligence has produced all or the main part of the creative elements of the recording: from the lead vocals to the main instruments, right through to music created entirely via a prompt. The term ‘AI-assisted’, on the other hand, will apply to works created predominantly by people, with the lead vocals and main instruments performed by humans, but in which technology has been used for certain expressive elements.
The distinction thus seeks to reconcile two requirements: to make entirely artificial content recognisable, whilst not equating works in which AI is used as a creative or technical tool with such content.
