8 min read

INEC'S ARTIFICIAL INTELLIGENCE REVOLUTION: WHO IS WATCHING THE MACHINE THAT WILL HELP POLICE NIGERIA'S 2027 ELECTIONS?

EXCLUSIVE INVESTIGATIVE REPORT

AI SYSTEMS TO SCREEN VOTER REGISTRATION, DETECT SUSPICIOUS PHOTOGRAPHS AND FLAG IRREGULARITIES — BUT WHERE IS THE FULL GOVERNANCE FRAMEWORK?

As INEC moves towards artificial intelligence-assisted election administration, questions emerge over transparency, voter privacy, false identification, human accountability and the danger of allowing an algorithm to influence citizens' democratic rights.

Nigeria may be approaching a new and largely unexplored chapter in the administration of its elections. The Independent National Electoral Commission (INEC) is moving towards greater use of Artificial Intelligence (AI), a technology that could help detect irregularities, identify suspicious registration records, improve logistics and strengthen the credibility of the electoral process.

But a fundamental question must be asked before the technology becomes deeply embedded in the machinery of the 2027 general elections:

Who supervises the artificial intelligence, who verifies its conclusions, and who takes responsibility when the machine gets it wrong?

An interview with an INEC official has brought this issue into sharper focus.

During the interview, the official described a system in which registration officers have individual login credentials to access an electronic platform. The system is being used to examine registration-related information, including photographs suspected of being artificially generated or otherwise irregular.

The official explained that the exercise was intended to identify irregularities across the states and bring questionable records to the attention of the appropriate authorities.

He also described other proposed applications, including an AI-powered question-and-answer service for citizens and the use of AI to improve the movement and distribution of electoral materials.

These are potentially useful innovations. However, the interview also exposes an important distinction: INEC's existing election-technology policy is not necessarily the same as a detailed, operational AI governance framework explaining how every AI application will work.

That distinction deserves national attention.

THE DOCUMENT INEC MUST MAKE PUBLIC

INEC already has a published document titled Policy on Election Technology Acquisition and Deployment (PETAD), dated 25 September 2025.

The policy is not imaginary, nor is it merely an internal proposal. It is an official document available on the Commission's website.

Its Section P specifically addresses Artificial Intelligence in elections.

It requires ethical and transparent use, compliance with applicable laws, rigorous testing, human oversight, bias mitigation, data security and continuous evaluation.

The policy expressly provides that AI must complement human decision-making rather than replace it. It also requires human verification mechanisms to prevent automated errors from affecting electoral processes and outcomes.

These are important safeguards. But principles on paper are only the beginning.

The public still needs to know the detailed operational rules governing the particular systems INEC intends to use.

The Commission's chairman, Professor Joash Amupitan, was reported on 6 October 2026 to have announced that INEC had established a dedicated Artificial Intelligence Division within its ICT Department. He also said an AI governance framework policy was being finalised, while automated AI auditing tools had been integrated into the results-verification process.

This raises an obvious accountability question: if the general PETAD policy already exists, what additional rules are contained in the AI governance framework that is still being finalised?

The distinction must be explained publicly.

WHEN AI FLAGS A NIGERIAN VOTER, WHAT HAPPENS NEXT?

This is perhaps the most sensitive aspect of the entire exercise.

Suppose a Nigerian citizen legitimately registers to vote. The citizen submits a photograph and other required information. An AI system examines the record and flags the photograph as suspicious.

What happens next?

Does the registration remain valid while the matter is investigated?

Is the individual invited to clarify the discrepancy?

Can a registration officer independently override an AI-generated alert?

Can a citizen appeal an adverse decision?

Is there a mechanism for correcting a mistaken identification?

These questions are not theoretical.

Artificial intelligence systems can make mistakes. Facial recognition and image-analysis systems can produce false matches, reject genuine photographs or misclassify images. Their accuracy can vary according to image quality, lighting, camera conditions, training data and the characteristics of the people being assessed.

A photograph flagged as suspicious is not proof that the person is fraudulent.

An AI-generated alert is an investigative lead, not a judicial finding.

INEC must therefore ensure that no Nigerian is denied registration, removed from the voters' register or otherwise deprived of electoral rights solely because an algorithm has raised an objection.

The final decision must remain with an identifiable, accountable human officer, operating under published rules and subject to review.

THE DANGER OF A TECHNOLOGY THAT THE PUBLIC CANNOT QUESTION

The Nigerian voter has a constitutional and democratic interest in knowing how the voters' register is compiled and maintained.

The voters' register is not the private property of INEC. It is a national democratic instrument.

If AI is used to identify duplicate registrations, suspicious photographs or questionable personal information, the Commission must explain the criteria employed.

What constitutes an irregular photograph?

What level of confidence is required before a record is flagged?

Does the system compare images against other photographs in the voters' register?

Does it use facial recognition, image manipulation detection, metadata analysis or a combination of technologies?

Who developed the software?

Where is the information stored?

Can an external technology provider access citizens' biometric or personal data?

How are system activities recorded and independently audited?

These are legitimate public-interest questions, not attempts to frustrate electoral reform.

Indeed, the more powerful the technology, the greater the obligation to explain how it is controlled.

BREAKING DEVELOPMENT: THE REGISTERING OFFICERS' ACCESS PROBLEM

The most significant development emerged on 30 October 2026, when INEC's Director of ICT, Dr Lawrence Bayode, reportedly explained on Arise Television that some registration officers had allegedly used their authorised login credentials to introduce questionable records into the voters' database.

According to reports of the interview, INEC's Automated Biometric Identification System (ABIS) helped identify questionable records and trace some of the irregularities.

This distinction is crucial: ABIS is a biometric identification system; it should not be confused with generative AI or an AI photograph detector.

Bayode also clarified that AI was being used for applications such as chatbots and election logistics, not for carrying out voter registration itself. TheCable +2

This changes the central investigative question.

The immediate problem is not simply whether AI can identify fake photographs. It is how authorised human operators were allegedly able to introduce questionable records into a national electoral database in the first place, and whether the existing controls were sufficient to prevent or detect that abuse.

If the officers had legitimate credentials, did the system have adequate safeguards against misuse?

Were their actions recorded in tamper-resistant audit logs?

Could supervisors identify which officer created, edited or approved each registration?

Were the credentials shared, compromised or deliberately misused?

And has INEC established individual responsibility for the irregularities?

These questions deserve answers before the Commission celebrates the technological sophistication of its electoral systems.

PETAD: A POLICY WITH TEETH, BUT DOES INEC ENFORCE IT?

INEC's PETAD policy contains provisions that should now be tested against the current controversy.

The document requires:

  • Role-based access control: Staff should have access only to functions authorised for their responsibilities.
  • Traceable system logs: Activities within the system should be recorded and auditable.
  • Data protection: Electoral information should be encrypted and protected.
  • Operational audits: Technology must undergo periodic examination.
  • Pilot testing: Systems must be tested in real electoral environments.
  • Human oversight: AI must not replace human decision-making.
  • Bias mitigation: AI models must be periodically reviewed and audited.

These are not optional public-relations promises. They are requirements contained in INEC's own published technology policy. inecnigeria.org

The central question is whether INEC has implemented these safeguards effectively across its national registration infrastructure.

A policy that cannot prevent the misuse of authorised credentials, or quickly identify and attribute suspicious transactions, is not enough.

THE LOGISTICS QUESTION: A USEFUL INNOVATION THAT NEEDS CONTROL

INEC's proposed use of AI chatbots and logistics support deserves consideration.

Nigeria's elections involve enormous logistical operations: ballot papers, result sheets, voting materials, BVAS devices, personnel deployment, transportation and distribution across thousands of locations.

An AI-assisted logistics platform could help officers obtain information, track materials, identify delays and coordinate distribution.

A chatbot could also answer frequently asked questions from citizens and election workers.

But the Commission must distinguish between administrative assistance and decision-making authority.

An AI chatbot must not invent election instructions. A logistics algorithm must not independently redirect sensitive electoral materials without authorised human approval.

Every movement of sensitive election materials should have an identifiable officer, a documented authorisation and a verifiable record.

THE SECOND FRONT: AI AND THE 2027 ELECTION RESULTS

There is another important dimension.

INEC Chairman Professor Joash Amupitan announced on 6 October that automated AI auditing tools had been integrated into the Commission's results-verification process, with human oversight.

On 14 October, he also publicly insisted that AI would not determine the outcome of the 2027 elections.

The distinction matters. AI may assist in detecting inconsistencies between scanned polling-unit result sheets and figures entered electronically, but it must not become an alternative returning officer.

An algorithm cannot substitute its statistical opinion for the legally prescribed electoral process. Punch Newspapers +1

The AI and Democracy Forum held in Abuja from 14 to 16 October also called for independent technology audits, transparent procedures and human verification of AI-generated alerts.

Its recommendations included publishing the governance framework and explaining how flagged results would be handled. Premium Times Nigeria

QUESTIONS INEC MUST ANSWER BEFORE 2027

The Commission should publicly provide clear answers to the following:

  • How many registration records have been identified as questionable?
  • How many suspected AI-generated photographs have been confirmed through forensic examination?
  • How many registration officers are under investigation?
  • Has INEC identified the individual officers responsible for introducing the questionable records?
  • What disciplinary or criminal proceedings have commenced?
  • What controls prevent unauthorised or improper use of registration officers' login credentials?
  • Are all changes to the voters' register recorded with timestamps and the identity of the officer responsible?
  • Has an independent cybersecurity audit been conducted?
  • What is the status of the separate AI governance framework announced by the chairman?
  • What independent testing will establish the reliability of AI-assisted results verification?
  • Can political parties and accredited observers inspect the procedures governing AI-generated alerts?
  • What remedy is available to a citizen wrongly affected by an automated system?

These questions are particularly urgent because INEC has indicated that it intends to complete the cleaning and verification of the preliminary voters' register before certifying the final register on 15 December 2026. TheCable +1

THE NATIONAL SECURITY DIMENSION: PROTECTING THE ELECTORAL DATABASE

Nigeria's voters' register is a national asset.

It contains sensitive personal information and forms the foundation of political participation.

An electoral database that can be manipulated by persons with legitimate access credentials raises concerns beyond election administration. It raises questions about cybersecurity, insider threats, institutional discipline and the protection of citizens' identities.

The investigation must establish whether the problem arose from deliberate misconduct, weak supervision, inadequate system design, compromised credentials or a combination of these factors.

INEC should not prematurely blame individual officers without presenting evidence. Equally, it should not allow institutional responsibility to disappear behind the allegation of misconduct by unidentified personnel.

The public deserves a transparent investigation.

OUR EDITORIAL POSITION: TECHNOLOGY MUST SERVE DEMOCRACY, NOT CONTROL IT

There is no reason Nigeria should reject artificial intelligence.

On the contrary, Nigeria should embrace technology that improves electoral credibility, reduces human error, strengthens logistics and protects the integrity of the democratic process.

But technology must never become an excuse for weakening accountability.

The responsibility for conducting credible elections remains with INEC. It cannot transfer that responsibility to software developers, technology vendors, biometric systems or artificial intelligence.

The Commission must publish its AI governance framework, subject its systems to independent audits, protect citizens' personal information and establish clear procedures for human review and appeal.

Most importantly, no Nigerian should lose the right to vote because a computer system made an error.

A machine may detect an irregularity. A machine may raise an alarm. A machine may recommend further investigation. But the responsibility to establish the truth and protect the citizen must remain with accountable human institutions.

Nigeria has suffered enough from electoral distrust. The 2027 elections must not become an experiment in technological opacity.

The country needs an election in which the voters trust the register, candidates trust the verification process, observers can examine the evidence and every electoral decision can withstand independent scrutiny.

INEC must prove that it controls its technology, rather than allowing its technology to control the credibility of Nigeria's democracy.

END.

Documents and reports for the investigation

Type at least two letters to search titles, summaries and topics.