The rapid advancement in AI technology has proved lucrative for certain sectors, the looming threat of AI for artists and creatives has always been a major concern. AI has evolved so rapidly, in fact, that generated imagery is virtually impossible to tell apart from actual photos and a person’s likeness, including their voice, can now be easily replicated.
This is of particular concern for the music industry. Songs using unauthorised simulations of popular artists’ voices have proven abundant and popular online. The most infamous case of this is Heart On My Sleeve. The track, which was released last year, uses the vocal likeness of pop artists Drake and The Weeknd. After appearing online, many believed the song to be legitimate. The track quickly gained millions of streams across platforms like Spotify and YouTube Music, without approval from either of the artists the song imitates.
Web based AI music generators like Suno and Udio have made access to voice simulating AI technology open to anyone with an internet connection. These apps allow users to create entire songs from pompts or recorded inputs. Udio, which was launched earlier this year, recently gained widespread attention when US producer Metro Boomin created BBL Drizzy – another track that featured Drake’s likeness that parodies his recent rap battle with Kendrick Lamar. Neither company has revealed their training data sources, leading to suspicions of unauthorised use of copyrighted materia Suno’s AI chief executive Mikey Shulman denies this, suggesting that Suno does not allow users to make reference to specific artists. Speaking to The Guardian, Shulman said that the “designed to generate completely new outputs, not to memorise and regurgitate pre-existing content.”
One of the biggest issues is how technology like Udio is being developed and made available for public use faster than it can be regulated. Current copyright laws do not take AI intro consideration, putting the intellectual property rights of recording artists at significant risk. A similar issue rocked the music industry at the start of the decade, as music consumption made the shift from analogue to digital sales. Streaming platforms also lack any ability to detect when an AI generated song that makes unauthorised use of an artist’s voice is being uploaded, meaning that anyone can create and upload fake songs by popular artists without their consent, and with no compensation to the artist.
These lack of regulations also stifle what could be a productive use of this technology for the music industry. Suno, which is available to use with a monthly subscription fee, has become popular amongst songwriters who pitch their songs to recording artists. Suno allows you to convert pre-recorded vocals to make it sound like any of the voices available on its servers. Songwriters have found this an effective method for submitting demos written with a particular artist in mind, and the artist and their label could essentially hear what they would sound like performing the song.
Now, major record labels and music industry groups are taking a stand, with several major groups suing Suno and Udio for copyright infringement. The lawsuits claim that Suno and Udio’s text-to-song services exploit protected works to create competing content that could potentially overwhelm and devalue original music. The plaintiffs, including Sony, Universal, and Warner, are seeking substantial damages per infringed work. The Recording Industry Association of America (RIAA) announced the lawsuits, with its CEO emphasising the unfairness of using artists’ work without consent or compensation.
Both Suno and Udio have gained significant funding and popularity, with Suno partnering with Microsoft and Udio being used by notable producers. Some artists have attempted to incorporate AI into their work in order to get ahead of any potential imitators. Last year, producer and pop singer Grimes made her voice available to be cloned by AI, and invited emerging producers to make use of it in their work in order to capitalise on the artist’s popularity and following. The project proved less than lucrative, however, with most of the music using the voice of Grimes AI failing to make any significant waves.
