INEC’s voter register cleanup raises questions about staff oversight

Nigeria’s electoral umpire, the Independent National Electoral Commission (INEC), says it has removed artificial intelligence-generated photographs from its voter register. But a more consequential part of the disclosure concerns the people who helped put them there: some of the commission’s own personnel.

With the 2027 elections approaching, the controversy raises questions that cannot be settled by deleting suspicious images. INEC needs to explain how questionable photographs entered the register, why its registration checks did not prevent their inclusion, and whether its newly announced image-capture restrictions will close the gap. The commission says it is replacing external webcam-based uploads with photographs captured directly through an approved application.

On November 11, 2026, the commission said it had identified 54 registration personnel across 10 states for possible disciplinary action over registration irregularities. This does not establish that all 54 uploaded synthetic photographs or committed criminal offences. It does, however, show that the inquiry extends beyond applicants submitting questionable information to officials responsible for protecting the registration process.

“But the ones involving our own staff, where it is our own staff that did so, that one is within our control,” Mohammed Kudu Haruna, INEC’s national commissioner for information and voter education, said during a PUNCH Media Foundation webinar on November 11. “They have been flagged,” he added. “Obviously, they are going to face disciplinary action.”

The allegations potentially extend beyond workplace misconduct. Section 23 of the Electoral Act 2026 criminalises knowingly transmitting materially false registration declarations and procuring the registration of fictitious or ineligible persons. Whether particular personnel committed those offences requires investigation; being flagged is not a finding of guilt. Haruna’s assurance of disciplinary action leaves a further question: will cases supported by evidence of criminal conduct also be pursued in court?

Beyond the photograph

The involvement of registration personnel changes the nature of the problem. Electoral bodies must anticipate inaccurate information from applicants. But when authorised staff allegedly circumvent registration protocols, the issue becomes whether the institution can enforce its own safeguards.

AI may add a technological dimension, but the investigation must address fundamental questions of institutional control. Who could upload or replace a photograph? What checks were required? Could one official make changes without independent review? And are there records showing who altered each entry?

Deleting suspicious images addresses the visible problem. Establishing how they entered the system is necessary to prevent a recurrence.

So far, the disclosures do not establish that an external cyberattack compromised the register or that manipulated photographs resulted in fraudulent votes. These possibilities should not be presented as confirmed findings.

What the numbers mean

The figures surrounding the cleanup also require careful interpretation. INEC Chairman Joash Amupitan said 1,608,660 invalid registrations had been removed. Separately, reports say the commission reviewed 268,000 contested cases, rejecting 217,722 and retaining 50,278. 

Published accounts have not clearly explained how these figures relate to one another, so they should not be added together automatically. Neither should the 1.6 million figure be described as the number of AI-generated voter records. 

The cleanup covered several irregularities, including duplicate entries and deficient registration records. Synthetic or manipulated photographs were one concern within that broader exercise.

The distinction matters because imprecise figures can turn an administrative problem into allegations that the available evidence does not support. At the same time, a large deletion figure does not explain what the commission detected, how it verified each category, or whether the underlying weaknesses have been addressed.

The 50,278 records retained after adjudication are equally important. Suspicion alone does not establish ineligibility. A credible cleanup must prevent invalid registrations while protecting eligible voters from wrongful exclusion.

Discipline is only the beginning

INEC says the register was published in preliminary form to allow public scrutiny and objections before final certification. That process can help identify errors before an election, but it should not replace safeguards that ought to operate during registration.

Both things can be true: public inspection may have helped expose irregularities, while their discovery points to weaknesses that the commission must address. INEC’s credibility will depend on its ability to explain what went wrong and how its internal controls are changing.

The proposed disciplinary action is a starting point, not a completed accountability process. The identified personnel must be allowed to respond to the allegations, and the commission must distinguish between mistakes, negligence and deliberate falsification. Disciplinary action should also not be confused with criminal prosecution, dismissal or conviction.

Beyond individual sanctions, Nigerians need to know what investigators found, which procedures failed and what corrective measures have been introduced. While personal information may require protection, INEC can still disclose its institutional findings.

The policy test ahead

INEC said in October it is finalising an AI policy, but the controversy raises questions about whether that framework will address only AI-generated campaign content or also the commission’s own systems, personnel and use of automated tools.

“We have that policy. Very soon, we will concretise it, and we will be able to share it with the media,” Haruna said at the November 11 webinar.

The timing is significant. With presidential and National Assembly elections scheduled for January 16, 2027, the policy was still awaiting publication roughly two months before polling. More importantly, the registration exercise had already exposed irregularities that the framework could have helped prevent.

A useful policy must go beyond broad warnings about artificial intelligence. It should clarify who can access and alter voter records, what verification is required before photographs are accepted, how suspicious submissions are escalated, and who independently reviews sensitive changes. It should also establish clear responsibilities for staff and supervisors, with decisions documented and open to audit.

The absence of a published standalone AI policy does not prove that INEC had no existing safeguards, nor does it mean earlier publication would necessarily have prevented the irregularities. The commission must explain whether controls were missing, inadequate or deliberately bypassed. Its own admission of staff involvement makes that distinction central.

Ultimately, Nigerians need more than assurances that suspicious photographs have been removed or that personnel will face discipline. They need evidence that the registration process is secure, that eligible voters will not be wrongly excluded and that those entrusted with the register are accountable.

The photographs can disappear quickly. The reason they were accepted must not disappear with them; otherwise, the next AI-generated image may be harder to detect, while the institutional weakness remains unchanged.

Source:

TechCabal