The intersection of artificial intelligence and the music industry has become a flashpoint of creative control, with established recording artists increasingly unwilling to let their life's work fuel technology they neither understand nor endorse. Over the past year, the three major record labels—Universal Music Group, Sony Music, and Warner Music Group—have begun striking deals with artificial intelligence startups, yet these arrangements proceed without securing explicit consent from the very musicians whose performances form the foundation of these AI systems. This disconnect between corporate strategy and artistic agency reveals fundamental tensions in how the music business values creators in an era of rapid technological change.
The core issue centres on a distinction often overlooked in licensing discussions: while major record labels own the rights to distribute recordings, they do not automatically control whether artists' voices and likenesses can be used for new technological applications. Madonna and her manager Guy Oseary made this position unambiguous in recent public statements, with Oseary declaring on the Tim Ferriss podcast that the global superstar categorically rejects AI training of her music regardless of financial incentives offered. Similarly, R&B artist SZA has been vocally critical, responding to revelations that her work appeared in standard training datasets by stating on Instagram that no explanation could justify the practice. These high-profile objections signal broader artist apprehension about surrendering control over creative identity to autonomous systems with unpredictable outcomes.
The financial stakes have sharpened focus on how the music industry navigates this transition. Stock prices for Universal, Warner, and Spotify have declined steeply as investor concerns mount regarding artificial intelligence's long-term impact on traditional streaming and licensing revenue models. To counter negative market sentiment, label executives have accelerated partnership announcements with AI music platforms including Udio and Suno Inc, both of which enable users to generate original compositions through text prompts. Warner and Universal signed agreements with these companies after initially pursuing litigation against them for copyright infringement—a strategic reversal that prioritises deal-making over enforcement. These announcements, however, remain largely devoid of artist participation details; executives claim widespread artist buy-in but refuse to name individual participants, creating credibility gaps that extend beyond investor relations into questions of authentic consent.
The mechanics of current deals expose how record labels leverage ownership of sound recordings without necessarily securing the additional permissions required for voice synthesis and personality replication. Universal's chief digital officer Michael Nash acknowledged on an analyst call that the company has conducted extensive conversations with thousands of artists and estates, implying many have agreed to participate in AI initiatives. Warner's chief executive Robert Kyncl similarly framed artist permission-gathering as an ongoing, complex process requiring systematic development of consent frameworks. Yet this framing—positioning permission-seeking as a future challenge rather than a prerequisite—suggests that label partnerships have already been inked before comprehensive artist agreements materialise. The practical consequence means artists discover their work powers new AI tools only after official announcements, leaving them limited leverage to negotiate terms.
Sony Music has adopted a more cautious posture, remaining engaged in active litigation against both Udio and Suno while approaching new deals with greater deliberation. This stance reflects recognition that the legal architecture governing AI training rights remains unsettled; rushing into arrangements risks establishing unfavorable precedents that could disadvantage artists seeking to establish market rates and protective clauses for voice synthesis applications. The regulatory and legal landscape across major markets including the European Union and potentially Malaysia's own copyright frameworks has not yet definitively addressed whether training AI on existing recordings constitutes infringement or requires separate compensation mechanisms distinct from traditional licensing fees.
The distinction between training AI models and enabling generative features introduces an additional layer of artist resistance. While recording companies may argue they can license their catalogues for model training under existing provisions, the prospect of allowing users to generate new compositions in specific artists' vocal styles or creative approaches meets with far greater hesitancy from the talent themselves. Artists perceive their voices as singular artistic assets whose commercial and reputational value depends on context and intentionality; allowing users to invoke a performer's name and style for arbitrary creative outputs—from novelty beach songs to potentially controversial or offensive material—represents a fundamental loss of creative control. This concern extends beyond financial compensation into existential questions about artistic identity and legacy.
The emerging framework that artists and their representatives are demanding includes explicit financial arrangements ensuring creators receive compensation when their voices or styles power commercial AI products, legal protections preventing unauthorised synthetic voice generation, and contractual language specifying permissible uses and retaining artist veto power over particularly sensitive applications. These requirements represent a significant departure from traditional music licensing, which typically separated performance rights from mechanical rights in relatively standardised ways. AI applications blur these categories, creating scenarios where an artist's entire vocal signature and stylistic fingerprint become raw material for generating theoretically infinite derivative works. Negotiating individual artist consent becomes labour-intensive for record labels accustomed to managing rights at the catalogue level.
For Southeast Asian and Malaysian music markets, this international conflict carries particular relevance as local artists increasingly establish regional and global profiles through digital platforms. Malaysian musicians who have built followings through streaming services and social media will eventually face similar pressures from international labels regarding AI training rights. The precedents established by negotiations between major Western artists and technology companies will likely influence terms offered to emerging regional talent. Additionally, as artificial intelligence music generation becomes more sophisticated and accessible, the competitive dynamics of professional music creation in Malaysia could shift; independent artists using AI tools to produce commercial music might displace session musicians and producers who currently generate steady income from recording work.
The resistance from established artists also reflects uncertainty about AI music's commercial viability and market adoption. Despite investor enthusiasm and label partnerships, fundamental questions remain about consumer demand for AI-generated music or synthetic voice performances. Will listeners accept algorithmically-generated compositions lacking human creative intention? Will synthetic voice performances in established artists' styles cannibalise demand for authentic recordings, or will they create new revenue streams through novelty and experimentation? These unanswered questions explain why even artists open to technological innovation remain reluctant to irrevocably commit their identities to systems whose long-term value proposition remains speculative.
The broader narrative reflects a creative industry grappling with technological disruption while attempting to preserve artist agency and financial sustainability. Unlike previous technological transitions in music—from vinyl to digital, from ownership to streaming—AI training involves questions of identity and voice replication that strike at the core of what constitutes artistic work. Musicians are not simply resisting change; they are demanding that technological advancement be implemented through frameworks that recognise their creative contributions, compensate them fairly, and preserve meaningful control over how their artistic identities are represented and exploited. Whether record labels, technology companies, and artists can collectively develop mutually acceptable agreements will largely determine how artificial intelligence reshapes music creation and consumption across global markets, including Southeast Asia.
