Apple is asking artists and record labels on its music streaming platform to voluntarily label songs that have been created using AI. The new “Transparency Tags” metadata system for Apple Music was announced yesterday in a newsletter to industry partners. worldwide music businessAnd it consists of four categories including track, composition, artwork and music video.
The track tag should be applied when “a physical part of a sound recording” has been generated by an AI tool, while the composition tag covers other AI-generated composition elements, such as song lyrics. The artwork tag applies to static or dynamic graphics, but only at the album level. For all other AI-generated visual content – whether standalone or bundled with an album – the music video tag should be applied. Multiple transparency tags can be used together for tasks that require disclosure of more than one of these.
In its newsletter, Apple says its new tags are a “solid first step” toward achieving industry-wide transparency around AI-generated music, and that labels and distributors should “take an active role in reporting when the content they distribute is created using AI.”
Apple Music’s tagging system follows other efforts by competing music streaming providers to protect authentic artists from spam and impersonation, and help make AI-generated music easier for users to identify. Spotify is developing a new metadata standard for AI music disclosure with DDEX – a music standards-setting organization that currently lists a senior Apple Music executive Nick Williamson as Board Member. Deezer also made its AI music detection tool, launched last year, available on other platforms in January, while Qobuz introduced own Proprietary AI detection system last week.
Unlike Deezer and Qobuz’s proactive detection systems, Apple Music’s transparency tags are entirely optional (for now) and put the responsibility for AI disclosure solely on record labels and music distributors rather than platforms. Apple even says that determining what qualifies as AI-generated music and visuals will be left to the discretion of content providers, “similar to genres, credits, and other metadata,” and that no AI use will be considered on works that providers have not tagged.
Integrity policies have not yet worked for other AI labeling solutions. Given the lack of enforcement around Apple Music’s tagging system, I’m struggling to see why producers and record labels would actually be motivated to use it.
