How generative AI undermines Kenyan music with data colonialism

From early phonograph recordings to digital streaming, each technological wave has reshaped how African musicians create and share their work. However, generative artificial intelligence presents challenges unlike anything that has come before, particularly for musicians in Kenya and across the African continent.

When algorithms meet ancestral timbre

Generative AI has fundamentally altered the global music landscape, transforming how musical works are made, distributed, and experienced. While the technology promises democratised access for creators everywhere, it simultaneously intensifies precarious conditions and sparks industry conflict, especially between artists in Europe, North America, and the Global North. For those in the Global South, these problems take a more destructive form—they are not merely economic displacement but merge with long-standing patterns of extraction and marginalisation.

This analysis focuses on the intersection of these forces within Kenya's music industry, examining how AI's algorithmic logic collides with what can be called the communal soul: the embodied oral epistemology, ancestral timbre, and sovereign intentionality of indigenous musical expression. The vast data sets that power modern machine learning often underrepresent or misrepresent African musical genres. This systemic bias is worsened by the common practice of training AI systems on copyrighted content harvested from the internet without proper authorisation or compensation, a situation that has sparked considerable legal and ethical debate. Such disruptive innovations represent not merely a recent development but rather the newest chapter in a long history of Africa's encounters with external forces that have frequently proven harmful.

From West African talking drums to the griots of the Mandinka Empire, music's value has derived from its social function, spiritual significance, and role as a vessel for collective memory. The talking drums of West Africa did not simply accompany other instruments; they served as the primary means of inter-village communication. Similarly, the griots functioned as the oral library of their history, preserving culture across generations. African music, typically transmitted orally, demonstrates a profound interconnection with social life. This orality-based epistemology, manifested and embodied through live performance, stands in stark contrast to the statistical logic underlying AI training. The latter relies on disembodied, decontextualised information drawn from fixed databases, which reflects Western epistemologies and risks flattening African traditions, divorcing them from their sociocultural contexts.

A dual crisis: legal and cultural

The collision between African music's embodied epistemics and AI's statistical logic creates what this research calls a dual crisis. The first crisis is legal, rooted in the failure of Eurocentric copyright regimes—including the Berne Convention, adopted in 1888 and most recently revised in 1979—to account for African musical ecosystems. These frameworks prioritise individual authorship, creating a legal loophole that permits the unpaid exploitation of communally owned, orally transmitted traditions. The second crisis is cultural: AI's capacity to generate music devalues human creativity and produces culturally flattened, context-free synthetic compositions. This phenomenon, termed digital orientalism, represents a form of digital neocolonialism that allows technology companies in the Global North to extract cultural material from the Global South without tangible compensation to communities.

This directly mirrors classical colonial activities, where colonial powers once extracted physical goods to fuel industrial production. Today, multinational corporations harvest immaterial cultural data to power AI development databases. The pattern of extracting raw materials from the Global South for profit-driven manufacturing in the Global North continues, with the material shifting from the physical to the digital realm.

Research approach and methods

This study employed a qualitative phenomenological approach to explore how AI, copyright law, and musical practice intersect within the Kenyan context. This methodological choice enables a comprehensive analysis of stakeholders' lived experiences caught between communal tradition and algorithmic extraction. Kenya was selected because of its reputation as the Silicon Savannah of East Africa, its thriving digital music economy, and its specific efforts to protect Traditional Cultural Expressions through the Protection of Traditional Knowledge and Cultural Expressions Act of 2016.

Semi-structured, in-depth interviews with twelve purposively selected participants were conducted between January and February 2026. The sample was stratified to include creatives, legal experts, and technologists, ensuring a multi-stakeholder perspective. Participants were recruited through professional networks and industry associations in Nairobi. Purposive sampling employed strict eligibility criteria to ensure data relevance and reliability.

The inclusion criteria required a minimum of four years of professional experience in Kenyan music, law, or technology sectors, as well as direct engagement with digital music production, intellectual property litigation, or generative AI development. Participants needed to be primarily based in Nairobi to ensure relevance to the country's central legislative and technological hub. Exclusion criteria removed hobbyists, artists working exclusively with analogue media who had no digital supply chain interaction, and stakeholders whose primary jurisdiction or market operations lay outside East Africa.

To protect participants from potential professional backlash regarding sensitive intellectual property and economic disclosures, all identities have been anonymised using alphanumeric codes (P1 through P12).

Document analysis complemented the interviews. A doctrinal legal study questioned the compatibility of Kenyan law with the ontology of AI-driven music. This included stress testing three sets of documents. First, the Copyright Act (Chapter 130) was examined: statutory definitions of material fixation and originality were compared against the fluid, oral character of Kenyan musical traditions to identify exclusionary triggers. Second, the Protection of Traditional Knowledge and Cultural Expressions Act was assessed using criteria of community ownership, with analysis focused on enforcement mechanisms to determine whether the Act enables communities to assert rights over training data or reverts to state-centric bureaucratic control. Third, the terms of service and data licensing agreements of prominent generative AI platforms (OpenAI and Suno AI) were audited. Asymmetrical clauses using standard open licensing to circumvent local customary restrictions were analysed through the lens of inequitable openness.

Controlled sonic analysis and algorithmic prompting

Researchers also performed controlled sonic analysis of outputs from two large commercial AI music generators: Suno AI and Udio. This assessed the aesthetic and cultural effects of generative AI on African music. During January–February 2026, targeted text-to-audio prompts were applied to these platforms, using descriptors of Kenyan and regional genres including a Kenyan Benga track with complex guitar lines, a traditional rhythm, and a Zimbabwean melody. Twenty generated tracks were produced—ten from each platform. Researchers analysed these outputs and cross-calibrated them against expert auditory feedback from music producers in the sample to determine rhythmic quantisation, melodic stereotyping, and the decontextualisation of ancestral timbre.

Thematic analysis using the six-phase framework of Braun and Clarke was applied to interpret qualitative data from interview transcripts. This began with extensive familiarisation, reading transcripts repeatedly to grasp each respondent's holistic story. Open coding followed, in which raw data points referring to loss of soul, unpaid labour, or community theft were synthesised into candidate themes. These codes were grouped through an iterative review process, revealing clear patterns in experience. For example, economic fears from technologists about market changes were collated into the theme of universal precarity, while creative practitioners' existential fears of erasure became the theme of ontological erasure.

These thematic conclusions were subjected to triangulation with findings from critical legal analysis to reveal the underlying structural glitch. This legal-ethnographic triangulation cross-referenced statutory failures with participants' lived experiences. For example, the legal ruling that unfixed works fall into the public domain under copyright law was directly overlaid onto narratives from classical instrumentalists, who explained they could do nothing when AI reproduced their unique, unrecorded beats. This confirmed the incompatibility theory that identifies tension between formal legal frameworks' material fixity and the informal reality of African creative practices that are fluid, oral, and communal.

Context from the literature

While the impacts of technology on music and the failures of intellectual property law are well documented, the specific intersection with AI—analysed through the lens of data colonialism and centred on African epistemologies—remains underexplored. The current anxiety around AI is the latest instance of technological apprehension facing African music. History reveals a consistent pattern of struggling with new technology, typically proceeding from initial fear to creative borrowing. This capacity has led to aesthetic reconstitution, where artists appropriate foreign technologies—such as electronic instruments used for Afro-rock or synthesisers in South African Kwaito—to create new culture.

The current moment differs significantly, as AI alignment risks epistemicide by imposing Western values that exclude Indigenous Knowledge Systems, forming a form of digital cultural colonialism. Consequently, the fear of cultural appropriation risks dominating creativity by enforcing rigid boundaries. This provides a critical counter-narrative to the ideology of technological determinism.

"Whereas creators in the North worry mostly about the loss of livelihood, African artists are dealing with the threat of ontological erasure—an elemental loss of cultural essence whereby indigenous knowledge is translated into statistical noise."

The emergence of AI has created a general atmosphere of precarity across the world's creative industries, but the specific threat differs significantly across geopolitical borders. In the Global North, where concerted advocacy and strong opposition from European and North American musicians dominates, anxiety manifests mainly through labour strikes and high-profile copyright lawsuits. Arguments focus primarily on economic displacement and the free use of copyrighted material. African creators face a threat that extends far beyond economics and approaches the existential.

What functions as a labour dispute in the North becomes digital colonialism in the South, where cultural resources are robbed without mutual value or appreciation. In this context, African music is not simply a repository of copyrightable information; it is a social archive threatened by a new type of epistemic violence.

The regulatory glitch

The glitch within the intersection of law, technology, and African music is not a technical error but a structural incompatibility between Western intellectual property frameworks and the ontology of African creative practice. Existing IP systems rely primarily on an individualistic approach to creation, which is fundamentally unequipped to defend expressions generated by communities. This involves a twofold failure: customary norms are underestimated, preventing communities from asserting their rights, while state-centric heritage laws often displace community control. This separation is exacerbated by global treaties such as the Berne Convention, which bases protection on material fixation—a concept inconsistent with the living archives of African oral tradition.

This introduces an ambiguity gap where existing laws struggle to balance Indigenous Knowledge Systems with the non-human agency of algorithmic output. This era represents a meta-crisis of generativity, where Western-centric standards threaten epistemicide, necessitating a shift toward context-dependent moral frameworks.

This legal trap is further complicated by international forces of data colonialism and the economic ineffectiveness of the training data supply chain, where the value of creative inputs is neither traced nor compensated. While recent litigation such as the New York Times lawsuit against OpenAI highlights the precarious nature of using copyrighted material as the basis for Retrieval-Augmented Generation models, the situation of African creators is unique. Standard licensing frameworks often reinforce exclusion and result in inequity by treating African dataset owners the same as well-resourced Global North entities. This injustice arises from treating unequally situated actors as equals, since standard open licences do not account for unequal infrastructure and capacity. This dynamic facilitates an extractive logic where cultural data is harvested without reciprocal value, effectively subsidising global AI development through the unremunerated labour of human creators.

The introduction of media technologies in Africa has a long historical context, which allows us to see AI not merely as a tool but as a social and cultural negotiation. African engagement with technology has historically been marked by agency; as Haupt (2008) observes, locally acquired instruments have been adapted to carry indigenous stories. In the past, radio, cassettes, and mobile phones served as spaces for negotiation in cultural production (Jedlowski et al., 2025). However, AI introduces a different paradigm through Machine Learning (ML). Unlike earlier technologies that captured or broadcast sound, ML models are trained on large datasets, effectively cannibalising the digital archive to generate new works. This creates a dual reality. On one hand, AI offers immense potential for innovation and the democratisation of production, as shown by successful examples like M-Pesa in Kenya. On the other, this dependence on the archive carries unspoken threats. Unbalanced training datasets risk being Eurocentric and digitally orientalist, leading to dehumanising and oversimplified representations of complex cultures (Odeke and Kirui, 2025). Furthermore, these gaps are worsened by infrastructure deficits and digital divides, threatening not only cultural integrity but also the economic survival of African creatives.

Case studies

Nigerian artist Eclipse Nkasi exemplifies a pragmatic adaptation paradigm in sonic innovation. Unlike visual artists who often create their own data, Nkasi has used commercially available options to produce his AI-generated album, Infinite Echoes. By applying AI to lyric writing and voice synthesis, he has bypassed the economic gatekeepers of the traditional studio system, cutting both production costs and barriers to entry (Umahi, 2025). Yet Nkasi maintains a human-in-the-loop approach; for him, emotional resonance and final structure must remain human responsibilities, even though AI supplies the raw materials, or digital clay. This example shows how African artists are repurposing extractive technologies to achieve economic sovereignty in an industry constrained by limited resources.

The Nigerian music industry also reveals the dualities of AI adoption. While small-scale artists like Nkasi feel liberated, larger organisations such as Mavin Records are testing AI's limits. Nkasiobi Chukwu of Mavin advocates for AI's democratising power, stating, "Any idea you have, you can make it a reality. That is radical" (Eleanya, 2025). Conversely, traditionalists argue that AI-generated Highlife or Juju songs often lack ancestral timbre—the spiritual resonance achieved through live performance. This tension exposes a divide between digital natives, who view music as content, and cultural purists, who see music as lineage (Eleanya, 2025).

In Kenya, the music production industry is marked by a rhythmic resistance approach. Interviews with producers of Benga and other genres reveal intentional subversion of AI model quantisation. P3 (Music Producer) described a method of sonic re-humanisation: AI-generated stems are de-quantised and then re-recorded with live percussion to restore the micro-timing details of African polyrhythms. By treating AI output as a sketch rather than a finished product, these Kenyan artists create a cultural shield, preventing algorithmic erosion of their heritage. This method mirrors the resistance of visual artists, focusing on the time and groove element of music, asserting that the soul of the rhythm cannot be automated.

Theoretical framework

To analyse the complex interplay between artificial intelligence, copyright law, and the African music industry, this paper uses the lens of digital colonialism. This theoretical prism is essential for examining the unequal power relations that shape interactions between Global North technology corporations and cultural producers in the Global South. The analysis rests on the idea that the pervasive extraction of data from human life constitutes a new colonial project (Kwet, 2019). This new social order builds on what Zuboff (2019) calls "surveillance capitalism"—an economic logic centred on appropriating human experience as free raw material for profit. Just as historical colonialism acquired land and raw materials, digital colonialism steals social interactions, behavioural patterns, and creative products, converting them into exploitable datasets (Couldry and Mejias, 2019). The commercial extraction of online information, including digitised musical heritage, to feed generative AI models is an expression of this reasoning today. These practices represent a novel form of data grab, where cultural artefacts are annexed, often against their communities' wishes, to create proprietary technologies that reinforce power disparities in the Global North, perpetuating algorithmic injustices (Birhane, 2021). This framework elevates the analysis beyond a critique of bias to a materialist examination of an extractive system, revealing how emerging technologies can recreate what Ndlovu-Gatsheni (2018) calls coloniality—the logic and persistence of colonial control—and exacerbate historical inequities in the digital era.

Findings

The doctrinal analysis of legal and policy documents reveals a multi-layered legal architecture that, overall, enables data colonialism. International structures enforce a hegemonic logic ill-suited to African cultural traditions; regional organisations offer a disjointed response; and new sui generis national laws provide a central but jurisdictionally limited mode of resistance.

The inadequacy of international frameworks

The basic tenets of international copyright law, as expressed by the Berne Convention, are largely inapplicable to most African musical traditions. The requirements for protection—originality, individual authorship, and fixation in a tangible form—systematically exclude a vast body of cultural expression that is oral, communally created, and evolutionary (Ficsor, 2005, n.d.; Netshitenzhe, 2013). P2 noted that "most of the copyright laws do focus mostly on individual ownership... this arrangement makes it difficult to protect traditional songs as they do not have a single author." This creates a dangerous legal vacuum. As P6, an IP Lawyer, stated, "you know, modern copyright laws focus mostly on the individual ownership and fixation. This makes it nearly impossible to protect African music such as Benga... AI companies exploit this exact loophole." Because African musical heritage often fails to meet the strict requirements of Western IP law, it is typically considered public domain. P1, a veteran traditional musician, expressed the resulting vulnerability: "when I see a producer actually generating an AI song, a song extracted from my community, it is confusing how the laws will deal with that..." This legal status allows such heritage to be integrated into the massive datasets that power generative AI models. This chronic problem is not a mere oversight; it reflects a legal construct that favours one form of creativity over another, leaving vulnerable communities open to exploitation.

The ambiguity of regional frameworks (ARIPO vs. OAPI)

The main intellectual property institutions in Africa, including the African Regional Intellectual Property Organisation (ARIPO) and the African Intellectual Property Organisation (OAPI), offer different protection modalities that hinder a cohesive response to AI and data security. ARIPO, which covers English-speaking countries, has a flexible designation system, but its scattered protection and reliance on local courts are unlikely to suffice for complex transnational AI violations (Mlambo, 2017). P9, Label Executive, highlighted the jurisdictional nightmare: "you know, the regional legal frameworks such as ARIPO are actually badly fragmented, very slow, and in most cases do not support African music... I wish we had a stronger continental cooperation and recognition of community ownership." Enhanced pan-African collaboration and acknowledgement of community ownership are essential. OAPI, in contrast, offers a single system among its Francophone member states, providing simplified enforcement within its bloc (Tran & Attorney, 2025). This lack of harmonisation between the two parallel systems leaves the continent vulnerable to the borderless operations of global tech companies. One respondent noted that the legal systems are divided, slow, and fail to reflect cultural realities, calling for more effective cooperation, recognition of community ownership, and artist support.

The promise and peril of sui generis legislation

In response to international law's inadequacies, several African nations have developed sui generis legal frameworks. For example, the Kenyan government enacted the Protection of Traditional Knowledge and Cultural Expressions Act, 2016. This Act aims to decolonise intellectual property law by granting communities exclusive rights over their traditional knowledge and cultural artefacts, including music and dance. It requires prior informed consent for commercial use and defines ownership collectively, transcending the individualistic Western copyright paradigm. Practitioners keenly feel the limits of localised legislation. P4, a Nyatiti player, observed with frustration: "we actually have the Traditional Knowledge Act here in Kenya, but it is shocking that the big tech companies in the developed economies are actually using our cultural artefacts for free. A local law cannot easily stop a global algorithm." One respondent emphasised the need for community education on their rights, simplified registration and support systems, strict enforcement of benefit-sharing regulations, and public pressure and advocacy to ensure corporate accountability, so the law can effectively protect Kenya's traditional music and culture. Another respondent noted that "big companies still use our (African) culture freely." Nevertheless, localised legislation limited to Kenya faces substantial enforcement difficulties, especially when data appropriation occurs internationally. To effectively combat data colonialism, a unified pan-African strategy is crucial. Such a framework, alongside national laws, would enable African nations to negotiate with global technology corporations on equal terms.

The emerging global response

In May 2024, WIPO established the Treaty on Intellectual Property, Genetic Resources and Associated Traditional Knowledge, which requires mandatory disclosure among patent candidates whose inventions involve genetic materials or related traditional knowledge (Noe, 2025; Syam and Correa, 2024). The direct impact on generative AI and music is relatively small, as there are no specific approaches to transparency. This is because the treaty focuses primarily on patents, not copyright, which is more relevant to music and data scraping (Noe, 2025). P9 noted that "nothing has really changed since 2023, when they passed global rules... by the way, my indigenous music is still vulnerable to unauthorised use, and companies profit while communities get nothing. If you ask, I will tell you we need action, not just treaties." The limitations of localised legislation are keenly felt by practitioners. P4, a Nyatiti player, observed with frustration: "you know here in Kenya we already have the Traditional Knowledge Act; I wonder why the big tech companies in the Global North still use our culture freely without any consent or compensation. Do you mean that the local laws cannot easily stop these foreign algorithms?" The treaty does not support the creation of new legal protections for traditional knowledge or cultural expressions like music. It also does not fully cover digital sequence information (DSI) or its derivatives, creating potential loopholes (Noe, 2025; Syam & Correa, 2024). However, a major hurdle remains the ratification and implementation of these new safeguards. Noe (2025) reports that, as of December 2024, 38 treaties had been signed but never ratified in Malawi. This has led to a prevailing belief that little has changed regarding the vulnerability of traditional music, reflecting broader weaknesses in established intellectual property regulations and the continued need for specialised legal tools. Syam and Correa (2024) conclude that the treaty is an important yet insufficient step toward equity.

The sound of algorithmic bias

The interplay of African music and AI is fraught with pitfalls of algorithmic bias and epistemic injustice, largely due to non-representative training sets and the constraints of existing AI models (Kirui et al., 2025, p. 20). When users input terms such as Benga, Afrobeats, or Zulu war chant, generative systems exhibit significant gaps, notably what can be called "Rhythmic Flattening." This is where the polyrhythms and overlapping timelines of many African musical cultures are reduced to generic 4/4 dance beats, lowering the tension and feel of the original styles as nuanced syncopations and multi-layered percussion are removed. P10, an AI Audio Developer, explained the technical mechanics: "the AI models are generally trained predominantly on Western 4/4 metres. When you prompt Suno for an African polyrhythm, for instance, the algorithm automatically quantises the rhythms to the conventional ones... Which, you know, of course, erases the human swing because its statistical baseline treats those syncopations as something like mathematical errors..." Another problem is Melodic and Harmonic Stereotyping. The AI tends to simplify into pentatonic scales or rudimentary melodic patterns more frequently, acting as a shorthand for "African-ness" rather than acknowledging the continent's diverse musical systems. For example, a "Kenyan Benga track" might generate a generic guitar tune lacking the characteristic complex lines and harmonic patterns of the genre. Lastly, and most significantly, is the Decontextualisation of Timbre. Chatterjee observed that "AI tools like Udio replicate melodies but strip them of cultural context" (2024, p. 1). Generative AI could mimic the sound of a Zimbabwean mbira, yet it remained an empty signifier, devoid of its connection to ancestral veneration, storytelling, and spiritual purpose within Shona culture. The sound was present, but its soul was absent.

Digital orientalism

Building on Edward Said's theory of Orientalism as a hegemonic discourse of representation (Said, 1978), Digital Orientalism has often resulted in the algorithmic production of cultural stereotypes. Digital Orientalism is defined as the process where AI systems, mainly trained on biased, incomplete, and decontextualised datasets, generate and perpetuate essentialised, exoticised, and often inaccurate representations of non-Western cultures (Mahoney, 2022). P2, an Afro-pop musician, described the stereotyping in AI outputs: "wherever you actually prompt Suno for an Afrobeat track, it just gives you the most generic, watered-down pentatonic melody. It sounds like a Hollywood movie's idea of Africa, not what is actually happening in the streets of Nairobi." This is an automated form of stereotyping that reinforces a monolithic and ahistorical view of Africa. Just as 19th-century European writers crafted a textual 'Orient' to suit their ideological purposes, 21st-century AI models and algorithms risk entrenching similar biases by producing a sonic Africa. P8, an academician, discussing the decontextualisation of timbre, noted that "AI can actually mimic the acoustic signature of a traditional instrument just perfectly, but it at the same time completely strips away the cultural context of the music." This AI-generated African sound may echo the biases of its developers and data sources rather than genuinely representing diverse African musical idioms.

Resistance and adaptation

Far from being passive victims, African artists are actively engaging with AI, formulating tactics of resistance, appropriation, and co-creation. Nigerian artist Malik Afegbua, for instance, uses AI as a co-creation instrument, actively shaping the final artistic outcome rather than letting the technology dictate it (Heugas, 2024). P3, a Music Producer, detailed a strategy of 'sonic re-humanisation': "I always use AI to generate stems for padding, but you know, they are not enough, I must always de-quantise them and add a layer, especially for traditional percussion over it." In other words, AI gives producers digital clay, but they must enforce a cultural firewall. The soul of the groove can never be automated; it must always be played. This viewpoint aligns with seeing AI as a creative partner. P12, a Tech Entrepreneur, while highlighting AI's potential for independent creators, noted that "AI can never be rejected, you know... it is now saving, for instance, upcoming Kenyan artists who cannot afford to record in a high-end studio in Nairobi..." AI extractive tools are being repurposed to achieve economic sovereignty for Kenyan artists. This includes decolonising AI tools through training that primarily overcomes biases in datasets, ensuring precise cultural representation as exemplified by Afegbua's "The Elder Series" (Heugas, 2024). P11, an AI researcher, pointed to structural solutions: "the goal towards this resistance is not just about tweaking the audio but actually building our own African AI models that understand our music." Moreover, initiatives such as the San Code of Research Ethics (2017) and the Maasai Intellectual Property Initiative (MIPI) demonstrate African communities' proactive assertion of data sovereignty and cultural preservation against potential exploitation by AI systems (Dugeri, 2024).

Comparing Digital Orientalism with efforts to decolonise the Digital Audio Workstation (DAW) reveals a crucial insight: in an AI-driven era, the soul of African music does not reside in a sound wave itself but emerges from the creator’s intentionality. Generic AI models produce sonically appealing yet soulless music, generating a statistical recombination of data devoid of lived experience. Van Schaik (2025) contends that listeners still believe computers cannot surpass humans in creating emotionally sensitive and stimulating music. Ultimately, what Nkasi and the Kenyan producers achieve can be called “soul”: an intentional, autonomous act of cultural self-representation, mediated by technology but not controlled by it. Consequently, the fight against soulless AI music reflects a broader struggle by artists to maintain control over narrative, context, and the means of cultural production.

Discussion

This section synthesises findings from legal research and semi-structured interviews to build an empirical argument on how generative AI affects African music. It interrogates the interplay among legal systems, cultural representation, and resistance, thus illuminating general applicability to postcolonial theory and related cultural contexts.

The vicious cycle and the virtuous resistance

Applying the data colonialism analytical framework, the current cycle of algorithmic extraction mirrors past colonial dispossession. Accordingly, artists’ digital resistance should be understood not merely as a technical workaround but as a necessary postcolonial praxis of reclaiming sovereign agency. The empirical evidence indicates persistent tension within the African music ecosystem. Hegemonic legal regimes have created structural incompatibilities that foster an environment enabling technology companies to undertake large-scale mining of cultural information. Rather than simply restating the mechanics of algorithmic bias, it is essential to consider the ontological implications of music. Algorithmic misrepresentations are not only economically harmful but also reinforce pre-existing systemic injustices (Kirui et al., 2025). Nonetheless, fieldwork showed that Kenyan artists participate actively in this cycle.

Sonic re-humanisation and other acts of Digital Resistance exemplify a meaningful counter-movement that adopts alienating technologies in an active, subversive, and decolonising manner, thereby achieving creative sovereignty.

Implications for digital colonialism theory

This study significantly extends digital colonialism theory by providing a contemporary and aesthetic case example. It demonstrates that the influence extends beyond economic or surveillance dimensions, reaching into the shaping of cultural memory and artistic expression. Cultural identity artefacts are the raw materials extracted. Tracing how African artists actively resist digital dispossession challenges deterministic accounts of technology. The creation of sovereign datasets and the deliberate sabotage of quantised algorithms represent concrete forms of data decolonisation. Such practices go beyond critique by constructing alternative technological infrastructures (Walter et al., 2020) that destabilise the extractive model of digital colonialism and restore narrative authority. These outcomes push digital colonialism theory beyond economic surveillance into the realm of systematic cultural memory theft and homogenisation, revealing how algorithms can extract the soul and ancestral timbre of indigenous music.

Broader implications

While the challenges that AI poses to intellectual property and ownership are universal (affecting every musician globally), the data-mining trends, algorithmic stereotyping, and legal deficiencies described here pose an acute threat to oral and communal traditions. Such vulnerabilities are likely to recur in other regions with strong indigenous cultures, particularly in Latin America, Southeast Asia, and among First Nations communities in the Global North. The central conflict lies between the extractive logic of generative AI and the imperative to protect cultural sovereignty. Fundamental questions persist: who owns culture? Who has the right to represent it legitimately? What measures can ensure that the democratising effects of a datafied world are equitably distributed to forge a fairer digital future? Ultimately, the postcolonial analytical lens reveals that the specific Kenyan resistance to generative AI mirrors a broader, more global struggle over algorithmic control, which risks preserving the ontological integrity of collective cultural practices worldwide.

Conclusion

This study investigated the twofold crisis that generative AI has introduced to the Kenyan music industry. This crisis spans the inadequacy of existing legal frameworks and the consequent risk of ontological erasure. These challenges are exposed as contemporary manifestations of data colonialism. The analysis shows that communal heritage lacks protection under Western-centred copyright laws, thereby enabling uncompensated data mining that fuels extractive algorithmic models. As a result, those models frequently produce culturally flattened musical forms that destroy the ancestral context of the art. Yet the research also predicts strong counter-movements. By deploying digital resistance strategies, Kenyan artists act as sovereign agents within a postcolonial ecosystem, repurposing these tools to decolonise the digital audio workstation. In the algorithmic paradigm, an artist’s sovereign agency, lived experience, and intent remain the decisive elements that imbue music with its nature.

To safeguard African digital heritage, this study recommends a multi-stakeholder approach. Policymakers—especially within the African Union (AU)—must develop pan-African sui generis law and communal data trusts. Technology companies should pursue radical transparency in data sourcing, adopt moral licensing, and invest strategically in cultural competence. Furthermore, an artistic empowerment initiative grounded in digital literacy will foster collective advocacy, enhance bargaining power, and shift the culture toward embracing the human-in-the-loop paradigm, which prioritises genuine creative production. Cross-continental and longitudinal studies remain urgently needed to examine the future, along with targeted research into developing a community-owned, open-source AI system trained on ethically curated, sovereign continental information.