Nigeria Is Adopting AI. But Who Is Governing the Data That Feeds It? – Chukwuebuka Korie

Nigeria has a Data Protection Act, an enforcement commission, and landmark digital legislation on the way. But according to regulators, investors, and independent researchers, a law alone cannot substitute for the culture of data accountability the country still lacks, and without it, AI will entrench the same inequalities it promises to solve.

Nigeria had become a recognised player in the global data privacy and digital governance space. It was a moment of institutional pride. The country has built enforcement infrastructure, issued compliance directives, launched investigations against multinational platforms, and joined the Global Privacy Assembly alongside 60 other data protection authorities worldwide. Yet in the same breath, “rapid innovation without safeguards could erode citizens’ rights and weaken public trust.” He urged regulators, technology firms, and other stakeholders to place ethics at the centre of AI deployment. “Trust,” he said, “is the currency of the digital age. Without it, even the most advanced technological solutions will struggle to achieve their full potential.” Those two statements, the declaration of progress and the warning against complacency, capture exactly where Nigeria stands on data governance in 2026. The architecture exists. The culture does not yet match it. And as artificial intelligence becomes more deeply embedded in Nigeria’s financial services, health systems, public administration, and daily life, the gap between the two is becoming increasingly consequential. “Data governance is no longer optional. It is enforceable. And in capital markets, trust is the oxygen of the system.

The economic stakes of poor AI governance are no longer theoretical. Observers describes this as one of the most pointed warnings on AI risk ever issued by a senior Nigerian financial institution leader. Weak governance of AI systems could trigger capital flight, distort markets, and erode investor confidence. Across the world, capital flows toward markets that demonstrate predictability, governance, and trust,” he said, according to a statement issued by the Exchange. “If AI systems are opaque, discriminatory, or vulnerable to breaches, that risk is priced into companies, sectors, and ultimately into the country itself,” As AI becomes embedded in credit scoring, trading strategies, and compliance monitoring, failure to embed accountability could destabilise financial institutions outright.

Chukwuebuka Korie, an AI researcher and technology career coach, argues that this warning reflects a broader pattern already visible in how international investors assess Nigerian digital businesses. “When a company cannot demonstrate how its AI systems make decisions or cannot show that those systems were trained on reliable, representative data, that uncertainty is a liability,” Korie says. “Investors price that liability. Regulators flag it. And citizens, eventually, bear the cost of it. The NDPA 2023 and the General Application and Implementation Directive (GAID) issued by the NDPC in 2025 together provide a substantive legal framework for data protection in Nigeria. Under the GAID, organisations classified as Data Controllers of Major Importance, those processing the personal data of more than 200 individuals within a six-month period, are required to file compliance audit returns, maintain documented data processing activities, and demonstrate through those audits that AI-driven processing adheres strictly to the law.

These are meaningful requirements. But according to experts consulted for this investigation, the deeper problem lies not in the absence of regulation but in the quality of the data that Nigerian AI systems are built upon in the first place. Many AI tools adopted in Nigeria are built on datasets that reflect foreign realities, producing, in his assessment, tools that “do not understand our languages, miss our context, and even reinforce bias against local users.” The absence of robust data governance frameworks leads to siloed, inconsistent, and error-prone datasets.” Without standardised protocols for data collection, cleaning, and validation, he argued, organisations across Nigeria struggle to prepare data that is ready for machine learning applications. The consequences are not merely technical. In sectors such as healthcare, agriculture, and financial services, where AI tools are being deployed to make decisions affecting millions of Nigerians, flawed input data produces flawed and sometimes discriminatory output. Research published in Frontiers in Research Metrics and Analytics in late 2024 supports this concern at a continental level. The study found that AI systems trained on biased data had produced discriminatory outcomes in critical sectors, including finance, healthcare, and law enforcement, across Africa. Among the documented effects: loan repayment algorithms exhibiting gender bias that resulted in lower approval rates for female borrowers, and credit scoring models trained on Western financial histories flagging Nigerian borrowers as high-risk based on behavioural patterns entirely unrelated to actual creditworthiness. “Most AI tools adopted in Nigeria are built on datasets that reflect foreign realities, producing tools that do not understand our languages, miss our context, and reinforce bias against local users.” Nigeria is home to more than 500 languages. Yet the large language models that power most AI tools deployed in the country were built overwhelmingly on English-language data scraped primarily from Western internet sources. According to research cited by the 2025 Stanford AI Index, global AI activity from model production to investment remains heavily concentrated in a handful of high-income countries, with Africa representing a marginal share of global training data.

The practical consequences of this imbalance are already being documented within Nigeria. Research found that virtual assistants and chatbots routinely struggle to recognise common Nigerian names. Content moderation algorithms, the report found, frequently flag legitimate Nigerian content as suspicious. Automated pricing systems have been observed applying location-based assumptions that result in Nigerian users being charged more for services than users in higher-income countries. In October 2025, community initiatives such as the Sokoto Digital Village were documented attempting to address the language exclusion problem from the ground up teaching coding and digital literacy to out-of-school children in Hausa, because AI tools built in English were, in the words of one digital rights expert interviewed by The Whistler, creating “systemic biases where non-English speakers are either excluded, poorly represented, or misunderstood.” At the same time, the African Next Voices project, funded by a $2.2 million grant from the Gates Foundation and reported in Nature in November 2025, was recording 9,000 hours of everyday African conversations across 18 languages, including Hausa and Yoruba an effort to build the localised training datasets that Nigeria’s AI ecosystem currently lacks. Nigeria’s own National AI Strategy, finalised in 2024, acknowledges this challenge. The strategy identifies the development of a multilingual large language model starting with Yoruba, Hausa, Igbo, and Nigerian-accented English as a priority. The model remains at an early stage of development. Many describes Nigeria as among the top 60 countries globally in AI readiness. We have welcomed this assessment while cautioning that such milestones “will remain symbolic unless backed by strategic investment in local data infrastructure.” The inclusion problem in its starkest terms. More than 60 million Nigerians, a figure exceeding the combined populations of Ghana and Rwanda remain offline. “You cannot talk about an AI-driven economy, “when young people in rural areas cannot even afford smartphones or data.” The implications for data governance are direct. AI systems reflect the data on which they are trained. Where those datasets are drawn predominantly from urban, educated, digitally connected Nigerians, because those are the Nigerians generating digital data, the systems built on them will serve urban, educated, connected Nigerians best. The 60 million who are offline are not only excluded from the benefits of AI; they are, in effect, invisible to it. Their healthcare needs, their agricultural patterns, their financial behaviours, and their languages are absent from the datasets shaping the decisions that will increasingly affect their lives.

Korie frames this as a governance failure as much as a technical one. “The question of who is represented in Nigeria’s data is not a secondary issue,” he says. “It is the primary issue. If AI is trained on data that excludes the majority of Nigerians, then AI will serve a minority of Nigerians. That is not an accident. It is a consequence of a governance choice or rather, the absence of one.” Federal Government, which already holds access to public data, must “take the first step” in building a centralised, representative national data pool from which both public and private sector AI systems can be trained. If AI is trained on data that excludes the majority of Nigerians, it will serve a minority of Nigerians. That is not an accident. It is the consequence of a governance choice.” Chukwuebuka Korie, AI Researcher. The argument made by Nigeria’s regulators, investors, and independent researchers is not that the country should slow its adoption of artificial intelligence. Nigeria’s digital economy is projected to reach $18.3 billion by 2026. The telecommunications sector alone contributed 9.2% of real GDP in the second quarter of 2025. The momentum is real, and the opportunity is significant. The argument, rather, is about foundation. Nigeria’s challenges in equitable AI adoption stem from inadequate policies, limited digital literacy, and reliance on foreign technologies that compromise data sovereignty. The study found that infrastructure problems, cost barriers, and sociocultural obstacles continue to cause unequal distribution of AI’s benefits, disproportionately affecting women, children, rural residents, and older people. Nigeria’s incoming legislative framework, including the National Digital Economy and E-Governance Bill expected to be enacted in the second quarter of 2026, promises to extend regulation into algorithmic transparency, AI risk classification, and mandatory audits. The NDPC has committed to using compliance audit returns as a benchmark for assessing responsible AI use in data processing activities, according to its March 2026 statement on joining the global joint declaration on AI-generated imagery. These are meaningful developments.

“AI cannot scale without the right backbone. From data centres to broadband penetration, we need to invest massively in the infrastructure that will power AI applications across sectors.” Legislation and infrastructure, Korie argues, must advance together. “Nigeria is building an AI economy on a data foundation that is fragmented, underrepresented, and largely ungoverned in practice,” he says. “The law can mandate accountability. But only investment in the foundation in clean data, representative datasets, local language models, and digital inclusion can make accountability real.” The speed of Nigeria’s AI adoption is not the risk. The risk is scaling innovation on an unstable base and discovering too late that the inequalities the technology was meant to reduce have been written into the algorithm from the start.