Nigeria’s Independent National Electoral Commission (INEC) is integrating artificial intelligence tools into its election-results verification process ahead of the 2027 general elections, but the policy intended to govern their use is still being finalised.

INEC chairman Joash Amupitan disclosed the plans at a press conference and national stakeholders’ meeting in Abuja on October 6, saying automated AI auditing tools were being integrated into the commission’s results-verification process. He said the system would operate under a five-pillar AI governance framework with mandatory human oversight.

The announcement has drawn attention to two related questions: what the technology will actually do, and how Nigerians will be able to scrutinise its use.

INEC’s stated position is that AI will assist election officials by identifying inconsistencies and potential security threats, not independently determine election outcomes. However, the commission has yet to publish the completed governance policy or provide detailed public information about the tools.

What INEC’s AI system is expected to do

The proposed system is intended to help officials identify discrepancies in election records.

One reported application involves digitising polling-unit result forms, including Form EC8A, and comparing the figures with relevant records held in the Bimodal Voter Accreditation System (BVAS), which is used to verify voters’ identities and accreditation.

If the system detects inconsistencies, it can flag them for human examination before results are processed or published. The distinction is important: an automated alert is not proof that a result has been manipulated, nor does it automatically justify changing a recorded figure.

INEC has also discussed using AI-based anomaly detection to monitor database traffic and strengthen cybersecurity around its digital platforms. These tools are intended to help identify suspicious activity and reduce the risk of unauthorised access or alteration.

The commission has referred to engagements with international technology organisations and election-support groups, including Microsoft, Google and the International Foundation for Electoral Systems (IFES). However, it has not publicly confirmed that any of these organisations is supplying the specific AI tools.

Important implementation details also remain unclear. Public reports have not established which products or models will be used, when the systems will operate, who will review alerts, or how the commission will measure their accuracy.

The five principles behind INEC’s AI governance framework

Amupitan said the proposed governance framework rests on five pillars:

  • Human oversight: Election officials retain responsibility for reviewing alerts and making decisions.

  • Transparency and accountability: The use of AI should be open to scrutiny, with responsibility for decisions clearly assigned.

  • Data protection and privacy: Electoral information must be handled in line with applicable privacy requirements.

  • Cybersecurity and risk management: The systems should be assessed for vulnerabilities, misuse and operational risks.

  • Public trust: The technology should support confidence in the electoral process rather than undermine it.

INEC has established a dedicated AI division within its information and communications technology department. But the governance policy intended to define how these principles will be applied was still being finalised at the October 6 announcement.

That leaves practical questions unanswered. For example, will the commission publish the criteria used to flag suspicious results? Will independent experts be allowed to test the system? Will political parties, observers and the public be able to understand how a disputed alert was resolved?

Without clear procedures, a general commitment to human oversight does not fully explain how accountability will work in practice.

Existing election technology already has legal and operational rules

AI is being introduced into an electoral system that already relies on digital tools, including BVAS and INEC’s Result Viewing Portal, known as IReV.

The Electoral Act 2026 provides legal recognition for key parts of this system and requires electronic transmission of polling-unit results to IReV. It also addresses situations in which transmission fails because of communication problems, providing for the manually completed Form EC8A to remain the primary record in the specified circumstances.

These provisions matter because AI auditing would not operate in isolation. Its findings would have to be considered alongside the official result forms, accreditation records, existing electoral procedures and applicable law.

The legal framework for electronic results transmission should not, however, be confused with a complete framework for AI. The commission’s separate governance policy is expected to explain how automated analysis will be managed and how its use will be supervised.

INEC has also cited the Nigeria Data Protection Act 2023 and the Cybercrimes Act 2024 as part of the wider legal environment relevant to data protection and cybersecurity.

As of the October 6 announcement, the commission had not given a public date for releasing its AI governance policy.

Osun election findings underline the importance of reliable source data

The accuracy of an AI auditing system depends partly on the quality of the records it examines. If source records are incomplete, inconsistent or affected by technical problems, automated analysis may identify discrepancies without necessarily explaining their cause.

The 2026 Osun governorship election offered a recent example of the operational challenges surrounding existing election technology.

Election observer group Yiaga Africa reported that BVAS functioned properly in 80% of the polling units it observed, while technical problems occurred in the remaining 20%. It also reported cases in which voters whose cards had been verified were unable to complete accreditation because their fingerprints or facial biometrics could not be authenticated.

Yiaga said the problems did not reach a level that undermined the election, but called for a comprehensive technical audit of BVAS ahead of the 2027 polls.

The organisation also raised concerns about interruptions affecting the upload of polling-unit results to IReV and called for greater transparency around the platform’s performance.

These findings do not demonstrate that INEC’s proposed AI tools will fail. They do show why technology testing, accurate source records, clear escalation procedures and effective human review are essential parts of electoral administration.

INEC has announced further preparations, including plans to test new BVAS devices and conduct mock accreditation exercises. The results of these exercises will be important indicators of operational readiness, although they would not by themselves establish the accuracy of the separate AI auditing system.

Political groups question the timing of the policy

INEC’s announcement prompted criticism from the Obidient Movement and the Yoruba Ronu Leadership Forum, according to reports published on October 7 and 8.

The Obidient Movement called for the AI plan to be suspended until a clear governance framework could be published and independently audited. It also called for transparent, traceable polling-unit results and public access to result forms through IReV.

The Yoruba Ronu Leadership Forum raised questions about who would have the final authority when an AI system flags a result and warned about the risks of manipulation and hacking.

These are positions expressed by political groups, not proof that the planned system has been compromised or that its use would necessarily undermine an election. Their concerns nevertheless highlight issues that a public governance policy could address: who is responsible for decisions, how errors can be challenged, and how independent scrutiny will work.

INEC’s position is that human oversight will remain mandatory. The outstanding question is how that commitment will be translated into published rules and verifiable procedures.

Deepfakes and election misinformation add another challenge

Beyond results verification, Amupitan has warned about the potential use of AI-generated voice cloning and deepfakes to spread false information during the election period.

Such content can make fabricated statements appear to come from candidates, election officials or other public figures. The risk is particularly significant when misleading clips circulate quickly before journalists or officials can verify their authenticity.

INEC has urged media organisations to verify election-related information with the commission before publishing it. Newsrooms, fact-checkers and social media users also have a role in checking the original source, date and context of viral content.

However, countering misinformation requires clear communication and transparent evidence. AI detection tools may assist the process, but their results should not automatically be treated as conclusive proof that a recording is genuine or fabricated.

What Nigerians should watch before 2027

Three developments will help clarify how INEC’s AI plan will work in practice.

First, publication of the governance framework. Nigerians need to know what the system can and cannot do, who is responsible for decisions, how personal data will be protected and what happens when the system produces an incorrect alert.

Second, testing and independent scrutiny. INEC’s planned BVAS exercises will offer information about existing election technology. Separate testing and independent assessment of the AI tools would help establish their accuracy, limitations and resilience before wider deployment.

Third, a transparent process for resolving discrepancies. The commission should explain how an alert is reviewed, what evidence is considered, who authorises any action and how affected parties can challenge a decision.

These measures would make it easier to distinguish between a system that merely generates alerts and one whose operation can be meaningfully scrutinised.

INEC’s proposed use of AI could help election officials identify inconsistencies in large volumes of electoral records and monitor suspicious activity across digital systems. But the value of that technology will depend on more than the ability to detect anomalies.

The commission has promised human oversight while acknowledging that its AI governance policy is still being finalised. That makes the publication of clear rules, credible testing and transparent accountability central to the next stage of the project.

AI should support the verification process, not replace evidence, established electoral procedures or the responsibility of election officials. For voters and observers, the crucial issue is whether the system’s findings can be checked and challenged through a clear, fair process.