A woman can be rejected for a loan in four seconds by an artificial intelligence (AI) system that never tells her what she did wrong, and Nigeria has no binding national rule requiring anyone to explain the decision.
Development Diaries reports that Professor Oluwatoyin Enikuomehin, Professor of Computer Science and Deputy Vice-Chancellor of Lagos State University, has called for urgent regulation of AI in Nigeria, warning of the risks posed by AI-driven decisions and systems.
He gave the warning on 25 August during the university’s 127th Inaugural Lecture at its main campus in Ojo, Lagos.
The warning comes as AI is already being used to determine who gets credit, who passes identity checks, who is flagged for fraud and who gets recommended for a job interview.
Nigeria has policies and laws touching these systems, but people affected by automated decisions still have little practical way to know when an algorithm has rejected them, what information influenced the decision or how to challenge the outcome.
The National Information Technology Development Agency (NITDA) has developed a national AI strategy, while the Nigeria Data Protection Act 2023 established the Nigeria Data Protection Commission (NDPC) and provides protections around personal data and automated processing.
What is missing is a binding, sector-wide requirement for operators of automated systems to explain decisions, test them for discriminatory outcomes and provide an accessible route to human review.
Digital lending shows the problem clearly, as loan applications can be assessed using device information, transaction histories, contact lists and behavioural data, allowing a system to reject an applicant almost instantly without explaining which information worked against them.
The Federal Competition and Consumer Protection Commission (FCCPC) has already taken action against digital lenders over how they treat borrowers, showing that regulators can intervene in the sector. But the automated systems determining who gets access to credit have received far less scrutiny.
The same problem appears in Nigeria’s identity systems. National Identification Number (NIN) checks, bank verification number (BVN) matching, and biometric authentication can produce false rejections that prevent people from accessing bank accounts, SIM registration, pensions and government programmes.
For example, a farmer with worn fingerprints or an older person whose fingerprints are difficult to capture can be rejected by a machine and left without a simple way to ask a human being to correct the decision.
The problem becomes wider when the data behind these systems is considered. Models built mainly on formal-sector records are more likely to work for people whose economic lives appear in those records than for Nigerians who depend on cash transactions, informal businesses and community savings schemes.
Rural women, for example, can be penalised by credit systems that rely heavily on formal transaction histories, while farmers, artisans, market porters and older people can face biometric failures because their fingerprints are worn or difficult to capture.
People with disabilities face another barrier when a service depends entirely on facial or fingerprint identification without an accessible alternative. Nigerians without smartphones or reliable internet access can also struggle to challenge an automated decision even when the law gives them a right to do so.
Responsibility for closing these gaps is spread across the NDPC, NITDA, the Central Bank of Nigeria (CBN), the FCCPC, and the National Identity Management Commission (NIMC). The National Assembly also has a role in creating a clear statutory framework for automated decisions across sectors.
Existing laws already provide some protection, with Section 37 of the 1999 constitution protecting citizens’ privacy, including their telephone conversations and correspondence, while Section 42 prohibits discrimination on several grounds, including ethnic group, place of origin and sex.
The Nigeria Data Protection Act provides rights over personal data and gives the commission enforcement powers. The Discrimination Against Persons with Disabilities (Prohibition) Act 2018 also requires accessibility in service provision, making alternative pathways important where biometric and digital systems exclude people with disabilities.
Nigeria’s obligations under the African Union Convention on Cyber Security and Personal Data Protection also require the protection of personal data and regulation of its processing.
Nigeria can begin to close the gap without waiting for a comprehensive AI law. The NDPC can start by requiring operators of automated decision systems to tell people when a decision affecting them was made by a machine and disclose the principal factors behind it within a fixed period.
That gives people something concrete to act on. Anyone denied credit, insurance, employment opportunities or a service through an automated system should request the basis of the decision in writing and keep a record of any refusal to explain it.
Complaints can then be lodged with the NDPC, giving the regulator evidence of patterns that individual applicants may not see.
Civil society groups can take that evidence further, with consumer and disability rights organisations testing digital lending and identity systems across different groups and publishing the results, while community organisations can document biometric failures by recording where they occurred, who was affected and what happened afterwards.
The pressure should also reach the people making the laws. Nigerians can ask their National Assembly representatives whether legislation governing AI and automated decision-making is before the chamber and publish the responses.
Regulators also need to move from broad mandates to specific safeguards. The NDPC should issue binding guidance on automated decision-making, including a right to explanation and a fixed response period, within 120 days. NITDA should publish a national AI framework for public consultation with clear requirements for testing discriminatory outcomes.
For digital lenders, the CBN should require licensed operators to disclose when automated underwriting is being used and provide applicants with a human review channel. NIMC should do the same on the identity side by publishing biometric authentication failure rates and providing an accessible manual alternative for people who cannot complete biometric verification.
In all, Nigerians should be able to know when a machine makes a decision about them, understand why it was made and ask a human being to review it.
