Resurrecting the Dead: The Rise of AI in South Korea’s Music Industry
How bringing back legendary voices like Kim Kwang-seok sparks thrilling nostalgia alongside complex legal and ethical battles.
South Korea's entertainment industry has pioneered a fascinating but controversial frontier: using artificial intelligence to bring deceased music legends back to the stage. This technological marvel was prominently showcased by the South Korean broadcaster SBS in its special program, "Competition of the Century: AI vs. Human." For the first time in the 25 years since folk rock superstar Kim Kwang-seok’s passing, audiences heard a newly generated performance in his exact voice. This milestone is part of a larger trend; music channel Mnet previously aired "One More Time" to honor classical musicians, and K-Pop sensation BTS performed alongside an AI-generated version of the late rocker Shin Hae-chul during a New Year’s Eve concert.
For fans, these AI-assisted concerts offer a deeply emotional and thrilling experience—a chance to hear a beloved, silenced voice sing once more. However, beneath the surface of this technological nostalgia lies a labyrinth of ethical and legal concerns. The primary question echoing through the music industry is highly complex: when a machine perfectly replicates a deceased person's voice, who truly owns the resulting new creation?
Under current legal frameworks, the ownership of an AI-generated performance is not assigned to a single entity. Instead, it is divided into three distinct legal categories:
The Right of Publicity:
A person’s natural voice is not covered by standard copyright laws, but rather by personality rights. In South Korea, these rights are inheritable. AI developers and broadcasting stations cannot legally generate and monetize a deceased artist's voice simply because they built the technology. They must obtain explicit consent from the surviving family members and financially compensate them for the use of the artist's likeness and identity.
The Copyright of the Final Song:
Copyright law fundamentally requires a work to be the expression of human thoughts and emotions. Therefore, an AI system cannot be recognized as a legal author. However, if human producers use the AI as a digital instrument—investing significant creative effort into prompting, arranging, and mixing the audio—the human engineers and producers own the copyright to that specific new musical arrangement, not the AI itself.
The Software and Model Rights:
The technology companies that build these sophisticated algorithms own the underlying code and the trained voice models. Yet, owning the software does not grant them automatic ownership over every song the AI generates, much like a guitar manufacturer does not own the rights to the songs played on their instruments.
As artificial intelligence continues to blur the line between life and death in the realm of art, the legal system will be forced to rapidly adapt. While developers own the code and producers own the master tracks, the fundamental essence of the performance—the human voice—remains safeguarded by the artist's estate. As this technology becomes increasingly accessible, establishing clear ethical boundaries and robust legal precedents will be crucial to ensure that honoring a musical legacy never crosses the line into exploitation.
