On 21 October 2026, at a press briefing in Abuja ahead of the International Cyber security Conference 2026, NITDA Director General Kashifu Abdullahi warned that AI-generated misinformation, deepfake videos and manipulated content on social media could threaten information integrity and sway public opinion before Nigeria's 2027 elections. He also said attackers increasingly exploit human trust through AI-powered social engineering. The briefing announced no new rules or enforcement measures, only awareness and verification guidance. That restraint is the right instinct, and Nigeria should keep to it.
The strongest case for tougher rules
The case for acting is serious. An SBM Intelligence report released on 1 July 2026 found that 66.5% of Nigerians say they are extremely or very concerned about AI-generated political misinformation. The same report says the Cybercrimes Act predates deepfake technology and that proposed AI legislation still lacks robust enforcement mechanisms. A synthetic audio clip released the night before a vote can spread faster than any correction, and a regulator waiting for perfect evidence may act too late. Those who want platform duties, takedown timelines and penalties are responding to a real gap, not inventing one.
Why the available rule-making tools fit badly
The tools Nigeria has used for online speech have a poor record against the problem they are now asked to solve.
The first is criminal speech law. Section 24 of the Cybercrimes Act, which criminalised certain online messages, was examined by the ECOWAS Court of Justice in a case brought by SERAP. In its judgment of 25 March 2022, the Court found the section was not in tandem with Article 9 of the African Charter on Human and Peoples' Rights and Article 19 of the ICCPR, and ordered Nigeria to amend it. Nigeria then amended the Act in 2024. When SERAP challenged section 24 of the amended Act (case ECW/CCJ/APP/03/25), the Court dismissed the claims. It found insufficient evidence that the provision's enactment or application violated the rights to freedom of expression, information or privacy, while confirming it can examine national laws for human rights violations. That outcome is not a clean bill of health for the provision. It shows how hard it is to prove harm from a vague speech offence, and that litigation is a slow remedy for an election-season problem.
The second is platform regulation by code. The Nigerian Communications Commission adopted its Internet Code of Practice on 13 February 2026, after a consultation that ran from 9 to 31 October 2025. The tracker lists a grace period from 15 May 2026 and entry into force on 13 August 2026. The NCC published Guidance Notes on the Internet Code of Practice 2026 alongside it. According to the briefing context for this piece, enforcement of the new platform rules is deferred under the minister's July order directing the NCC, NITDA and NDPC to pause. We could not retrieve the order itself, so treat that detail as reported, not independently confirmed.
What the NITDA approach gets right
A warning paired with verification guidance does three things a mandate would not.
First, it targets the actual failure. Deepfakes succeed when audiences cannot cheaply check a claim. Public guidance, provenance habits and fast reporting channels act on that directly. A takedown rule acts only after content is flagged, and it asks platforms to judge political truth at scale. Platforms that fear liability usually over-remove, and in an election the removed material is often legitimate criticism of those in power.
Second, it avoids giving a single agency a new lever over political speech months before a vote. Any power to order removal of "misinformation" will be used by whoever holds office. Nigeria's own experience with section 24 is the cautionary tale.
Third, it keeps the regulatory picture legible. Three bodies (NCC, NITDA and NDPC) have overlapping claims over platforms, and the pause exists because those claims have not been reconciled. Adding an election-specific deepfake regime on top would multiply uncertainty for the platforms and startups that Nigeria wants to build on.
A narrower agenda for the next twelve months
Caution is not inaction. A proportionate programme would include:
- Fraud and impersonation first. Abdullahi's point about AI-powered social engineering is a fraud problem. Existing fraud and identity-theft offences can be applied to synthetic impersonation without any new speech category.
- Transparency over removal. Ask large platforms to publish election-period reports on political AI content, labelling practices and takedown requests, including government requests.
- Narrow, judicial remedies. Where a deepfake impersonates a named candidate or electoral official, a court-ordered, time-limited remedy is more defensible than an administrative takedown.
- Resolve the jurisdiction question. Before the pause is lifted, the minister should publish how NCC, NITDA and NDPC will divide authority, so that platforms face one coherent set of obligations.
- Invest in verification capacity. Support independent fact-checkers and the electoral commission's own rapid-rebuttal channels, with published methodology and no government veto.
What to watch
The risk is not that NITDA spoke. It is that a public alarm, repeated through the campaign, becomes the justification for rushed rules after the first viral fake. Regulators should decide now what evidence would justify new obligations, and commit to publishing that evidence. If a measure cannot be tied to a specific, documented harm and a remedy that courts can review, it should wait. Nigeria has a rare chance to show that an election can be defended with transparency and verification rather than speech controls, and the 21 October briefing, whatever its limits, did not close that door.