By Dr Joachim Ezeji, PhD, MBA, PMP
The proposed deployment of artificial intelligence (AI) by Nigeria’s Independent National Electoral Commission (INEC) raises important questions about electoral integrity, transparency, accountability and public trust.
No doubt, AI can strengthen election administration and help detect irregularities, but it can also introduce new avenues for electoral manipulation. The outcome depends less on the technology itself than on how it is designed, governed, independently audited and integrated into Nigeria’s electoral processes.
There are three distinct risks to consider:
1. Manipulation using AI: Deepfakes, misinformation, voter suppression and targeted disinformation.
2. Manipulation of AI systems: Biased data, compromised algorithms, cyberattacks and unauthorised changes to election records.
3. Manipulation concealed by AI: Using technically complex systems to obscure decisions, frustrate scrutiny or create an appearance of credibility without adequate evidence.
The critical principle is that AI should assist election officials and independent observers, not become an unaccountable authority over votes or election results.
Against this background are the vituperations of the garrulous Minister of the Federal Capital Territory (FCT), Mr Nyesom Wike, which are becoming so unsettling to ignore. The most stupefying is his allocation of votes for an election that is yet to be conducted. For the umpteenth time, Mr Wike has said that his principal, Mr Tinubu, will win both Rivers and FCT ballots, in addition to other states.
But in making this wild claim, he has woefully kept silent on how he plans to achieve the feat. However, for a man who has never won a fairly contested election, the allegation that he triumphs by writing and sneaking in results and using the corrupt Nigerian judiciary to validate them has become timely for review.
Realising the pyrrhic task ahead of the APC in the January 2027 election, Wike is bent on having his nefarious way by all means. Notably, the APC has never won a presidential election in Abuja. The victory it recorded in Rivers State in 2023, from all evidence that emerged afterwards, showed that it was crooked/rigged through Mr Wike’s means.
It is in this regard that the plan by the Independent National Electoral Commission to deploy Artificial Intelligence (AI) for the election becomes manifestly disturbing.
The reason and approach behind this plan remain unclear. However, there appears to be a rigging masterstroke by the APC. This seemingly is bolstering confidence among APC stalwarts, particularly Mr Wike.
Already, Nigerians do not trust INEC, and this seems not to bother the commission. INEC has, to date, not explained the so-called glitch of 2023. Despite all its promises in 2023, the commission failed to live up to expectations. Now, to repeat the 2023 scenario in 2027, INEC has told a bewildered nation that AI will be deployed. This is probably a forewarning that the election would be rigged using emerging technologies such as AI and robotics.
The game plan runs thus: Rig the election, announce fake results, call out the military and police to maintain order, and ask opponents to go to already compromised courts. The implication is that unelected APC candidates are declared winners and imposed on the nation. If anyone disagrees, they are told to go to court.
INEC’s announcements describe intended capabilities and safeguards. They do not, by themselves, demonstrate that the systems have passed independent security and accuracy testing. This should worry every patriotic and keen observer.
For example, a particularly important danger is the possibility of false confidence in automated verification. An AI system might correctly establish that the figures on a polling-unit result sheet add up without establishing that those figures faithfully represent the votes cast. Similarly, an algorithm could identify unusual voting patterns, but statistical anomalies alone are not proof of fraud. They require investigation against original records and other evidence.
There is also the problem of deepfakes. Fabricated videos, cloned voices and forged announcements can mislead voters, while widespread awareness of deepfakes can cause people to doubt authentic evidence. INEC itself has acknowledged this concern but is yet to address it convincingly.
While we await INEC’s safeguards against these risks, it is important to emphasise that the risks can be substantially reduced through layered controls, although they cannot be completely eliminated. No credible percentage reduction can be stated without knowing the actual systems, their threat models, test results and operational arrangements.
To strengthen confidence and assure the Nigerian public that it means well, the following recommendations are necessary:
1. Independent technical assessment: INEC should commission technically independent assessments covering cybersecurity, algorithmic accuracy, data integrity, biometric error rates, system resilience and potential bias. Testing should include realistic election-day conditions, including poor connectivity, device failure and high transaction volumes.
2. Human oversight: INEC should ensure that AI is used only to identify potential problems and not independently reject votes, alter results or determine winners. Every significant automated finding should be reviewable by an authorised official, with documented reasons and an opportunity for appropriate challenge.
3. Verifiable chain of evidence: INEC should ensure an independently verifiable chain of evidence from polling units to the final declaration. This is, in fact, the strongest safeguard. It should cover:
a. Polling-unit results, signed original records and accredited voter counts.
b. Authenticated digital transmission with digital signatures, timestamps and integrity checks.
c. AI-assisted verification that flags anomalies without silently changing records.
d. Human collation and independent reconciliation.
e. Original records checked against transmitted and collated figures.
f. Publicly auditable declared results with traceable evidence, documented corrections and lawful review.
This approach ensures that AI does not become the sole source of electoral truth.
4. Independent technical oversight: INEC should encourage and ensure an independent technical oversight mechanism involving electoral administrators, political party representatives, cybersecurity specialists, academic institutions, civil society organisations and election observers. Such a mechanism should have meaningful access to system documentation, test findings, incident records and audit evidence, subject to legitimate security and privacy protections.
5. Biometric safeguards: INEC should note that AI-assisted biometric matching requires particular safeguards because errors can affect voting rights and breaches can expose sensitive personal information. Therefore, there should be strict access controls, independent testing for false matches, data-retention limits, incident reporting and accessible correction procedures for affected voters.
6. Public communication: INEC should explain what AI is doing, what it is not doing, who supervises it and how errors can be challenged. This is essential because unexplained automation can generate suspicion even when the technology operates correctly.
Before the 2027 election, Nigerians should insist on an AI assurance framework that clearly distinguishes between an AI system that merely produces a result and an electoral process that can demonstrate how the result was obtained and verified.
This can be achieved by adhering to the following six requirements:
i. Transparency: Built on a public inventory of AI systems and their permitted functions.
ii. Independent assurance: Based on summary findings of external security and accuracy audits.
iii. Human oversight: Anchored on documented procedures for reviewing automated findings.
iv. Data integrity: Based on traceable records and independently verifiable results.
v. Accountability: Anchored on clear responsibility for failures, corrections and incidents.
vi. Public scrutiny: Based on meaningful access for observers and lawful challenges.
Finally, the central question is not whether INEC should use AI, but whether its deployment will be transparent, independently verifiable, legally accountable and resistant to interference.
The safeguards announced by INEC, including human oversight and automated anomaly detection, address important concerns. However, independent testing and publicly demonstrable controls will be necessary to establish whether those safeguards are effective.
Three principles should govern the deployment:
First, AI must not replace the legal and evidentiary foundations of election results. Original electoral records and lawful collation procedures must remain authoritative.
Second, AI must not operate as a black box. Its findings, limitations and effects on electoral decisions must be explainable and open to appropriate independent scrutiny.
Third, technology cannot substitute for institutional integrity. Electoral credibility also depends on impartial administration, security, enforcement of electoral laws, transparency and effective dispute resolution.
Nigeria can use AI to improve the administration and verification of its elections. But the strongest protection against manipulation will come from combining technology with independent oversight, traceable evidence and enforceable accountability.
That is the standard against which INEC’s 2027 AI deployment should be assessed. This is very important to stop people like Nyesom Wike, who are hell-bent on compromising the 2027 elections.