The Independent National Electoral Commission has created a dedicated Artificial Intelligence Division inside its ICT Department to catch irregularities in election results and push back against disinformation before the 2027 general elections.
The commission’s chairman, Professor Joash Amupitan, announced the move on Tuesday at a World Press Conference and National Stakeholders’ Meeting in Abuja, held to mark exactly 100 days to the first ballot. He said automated AI auditing tools had already been built into the results verification process.
According to Amupitan, the tools run under a five-pillar AI governance framework that keeps human officials in charge of every decision. “Automated AI auditing tools have been integrated into our results verification process, under a five-pillar AI governance framework, operating under mandatory human oversight,” he said. He added that a full AI governance policy was being finalised, and that anomaly detection tools were being deployed to monitor database traffic and guard the commission’s public portals against intrusion or unauthorised changes.
The new division is a direct response to a threat the commission now ranks among the most serious facing the polls. Amupitan said generative AI, voice cloning and deepfake video had made it cheap to fabricate the voices of election officials, forge result sheets and mimic official letterheads, then circulate them faster than they can be checked. He named fake videos of officials announcing results, cloned audio directing the manipulation of voting devices, and forged letters announcing postponements among the scenarios the commission is preparing for.
His deeper worry was not any one falsehood but the erosion of trust that follows. “When deepfakes render lies believable, real truth is treated with scepticism, and cynicism replaces civil engagement,” he said. The risk, in his account, is that once convincing fakes spread widely, Nigerians begin to doubt even genuine information from legitimate authorities.
Amupitan placed a clear share of that burden on the press. He urged media organisations to set up specialised fact-checking desks, use forensic image-analysis tools, and verify election information with INEC before publication. He also drew a firm line on results, cautioning newsrooms against publishing privately collated figures or premature declarations. Under the Electoral Act, he stressed, the authority to declare results rests exclusively with the commission. “Resist the declaration of privately collated results,” he said.
The AI push sits within a wider technology drive for 2027. The commission said it had acquired 57,000 additional Bimodal Voter Accreditation System devices and completed a nationwide inventory audit covering activation and GPS-tracking verification. A six-day stress test of all configured devices is scheduled for 23 to 28 October, followed by a nationwide mock accreditation and election simulation in November to test staff, hardware and network systems under near election-day conditions. A situation room is also planned for real-time monitoring before, during and after the vote.
Context explains the urgency. The presidential and National Assembly elections are set for 16 January 2027, with governorship and state assembly polls on 6 February, which leaves the commission in what Amupitan described as a 100-day sprint. His caution to editors is not new either. Last month, at the All Nigeria Editors Conference in Enugu, he made a near-identical appeal for fact-checking desks and forensic verification, signalling that the media’s role in the information fight has become a settled theme of INEC’s preparations rather than a one-off remark.
The commission is also leaning on partnerships it has not fully detailed. Amupitan said INEC was working with international technology firms and partners to integrate its anomaly-detection tools, though the identities of those partners and the terms of the arrangements were not disclosed. The AI governance policy that would set out the rules for all of this is still being finalised, so the framework now in use rests on principles the commission has stated publicly rather than on a published document.
Nigeria’s electoral technology has drawn intense scrutiny before. The handling of result transmission in the 2023 general elections, particularly the delayed upload of presidential results to the INEC viewing portal, became a central grievance in the disputes that followed and shaped much of the public argument over the credibility of that exercise. By building verification checks into the results process this time and pressing the media to slow down before publishing, the commission appears to be trying to close the gap between the moment results are collated and the moment they are trusted.
What remains to be seen is how the tools perform under pressure. Automated checks are only as good as the data and oversight behind them, and the commission has given few specifics on how the AI systems will function, what they will flag, or how disputes over flagged results will be resolved. The stress test in October and the mock exercise in November will offer the first real measure of whether the technology works as intended. For now, the division exists, the policy that will govern it does not yet, and the 2027 timetable leaves little room to get either wrong.