Deepfakes Threaten Voter Trust Ahead of 2027
Fadekemi Ajakaiye
Artificial intelligence-generated deepfakes could become a growing threat to voter confidence in Nigeria’s 2027 elections, making it harder for citizens to distinguish authentic political communication from fabricated images, videos and audio, a cybersecurity and artificial intelligence governance expert has warned.
The risk extends beyond the production of false political material. As generative AI makes synthetic content cheaper and easier to create, the greater threat may be the erosion of trust in genuine evidence, leaving voters increasingly uncertain about what they can believe.
Toluwani Akinniyi, a cybersecurity and artificial intelligence governance expert, said Nigeria’s challenge ahead of the election was therefore not simply the spread of misinformation but the growing difficulty of establishing whether political content circulating online was authentic.
Generative AI can now produce convincing images, clone voices, manipulate videos and generate synthetic material at a scale that previously required specialist expertise and considerable resources. That lowers the barrier for political actors and others seeking to influence public opinion.
“For Nigeria, this presents a serious concern ahead of the 2027 general elections,” Akinniyi said.
Nigeria is not new to politically manipulated information. During the 2015 presidential election, Cambridge Analytica was reported to have been involved in a campaign supporting then-President Goodluck Jonathan, while subsequent investigations documented the use of divisive material intended to influence perceptions of his challenger, Muhammadu Buhari.
By the 2023 election, the rapid expansion of social media had created a much faster distribution network for political misinformation, including false election results, coordinated accounts and manipulated content designed to create misleading impressions of public support.
A doctored video involving Hollywood actors Idris Elba and Matthew McConaughey illustrated how easily authentic material could be repurposed for political purposes. The video appeared to show the actors endorsing Peter Obi, but AFP Fact Check traced the footage to genuine WIRED interviews and found that the original material had been edited to insert a political message.
The episode was not an AI-generated deepfake, but it demonstrated that the manipulation of digital political content predates the latest wave of generative AI.
“What has changed is the technology available to carry it out,” Akinniyi said, adding that generative AI could make political manipulation faster, cheaper, more accessible and easier to reproduce at scale.
There are already indications that synthetic political material is entering Nigeria’s electoral information environment ahead of 2027.
In January 2026, Florence Ajimobi, former first lady of Oyo State, denied a viral video purportedly showing her making controversial comments about the 2027 governorship election. Her office said a forensic review found the footage to be AI-generated, including manipulated lip movements and cloned voice patterns.
During the June 2026 governorship election in Ekiti State, CDD Fact Check also identified an AI-generated image that was falsely presented as showing voters waiting to sell their votes.
Neither incident suggests that Nigeria’s 2027 election will necessarily be overwhelmed by AI-generated material. But they demonstrate how synthetic political content is beginning to enter an information ecosystem already marked by intense political competition and widespread social media use.
The more consequential threat, Akinniyi said, could arise when synthetic content is combined with targeted distribution and coordinated amplification.
A fabricated video of a candidate, for example, could be accompanied by false headlines, AI-generated photographs and coordinated social media posts before being amplified by influencers and distributed through private messaging groups. The resulting campaign could give fabricated material the appearance of widespread public acceptance before fact-checkers have time to establish what is genuine.
That creates a second-order problem known as the “liar’s dividend”. As synthetic media becomes more common, people can increasingly dismiss genuine evidence by simply claiming that it has been generated or manipulated by AI.
A politician genuinely recorded making a controversial statement could argue that the recording was synthetic, while authentic photographs or audio could similarly be dismissed as fabricated.
“The danger, therefore, extends beyond the creation of false information. It is the possibility of gradually eroding confidence in authentic information as well,” Akinniyi said.
The implications extend beyond content moderation to the use of personal data in political campaigning. Campaigns can use demographic, behavioural, location and preference data to identify groups of voters and tailor messages to them, raising questions about how personal information is acquired, processed and used to influence political choices.
In April 2026, the Nigeria Data Protection Commission and the Independent National Electoral Commission established a joint working group to strengthen data protection and privacy within the electoral system.
The NDPC said INEC held the personal data of about 94 million Nigerians and stressed the importance of responsible data processing by political parties, civil society organisations, third parties and vendors.
For election regulators, the question is therefore becoming broader than whether a particular political message was created by AI. It also concerns what data was used to identify its intended audience, how that audience was selected and who should be held accountable for its influence.
Yet the same technology that creates new vulnerabilities could also strengthen the defences around elections.
Before the 2023 election, Full Fact’s AI-supported tools helped fact-checking organisations identify and review claims at a scale that would have been difficult through manual monitoring alone. Poynter reported that the technology increased the number of claims that could be identified and reviewed from about 100 a day to as many as 100,000.
That kind of capacity could become increasingly important in 2027, when fact-checkers are likely to face an information environment spanning social networks, video platforms, private messaging services and increasingly sophisticated synthetic media.
Human verification will nevertheless remain critical, Akinniyi said, particularly given Nigeria’s linguistic and cultural diversity, which can make automated detection less reliable across different contexts.
He called for a broader election information-integrity framework involving INEC, the NDPC, political parties, technology platforms, media organisations, fact-checkers and relevant security agencies.
Political parties, he said, should establish clear rules governing AI-generated campaign material and impersonation, while INEC should strengthen systems for identifying and correcting false information that falsely appears to originate from the electoral commission.
The NDPC should also extend its scrutiny of political campaigning to voter profiling, third-party data and targeted political advertising, while technology platforms should establish stronger local mechanisms for responding to election-related synthetic media.
Media organisations will have to strengthen their own verification procedures before publishing political images, videos and recordings, while voters will increasingly need to question the provenance of content, seek independent confirmation and verify claims before sharing them.
Nigeria still has time to prepare for the technological risks likely to accompany the 2027 election, Akinniyi said. The challenge is to treat the information environment as part of election infrastructure rather than as a separate communications problem.
“Protecting an election is no longer only about securing the ballot box, voting technology or transmission system. It also requires protecting the information environment in which voters make decision,” Akinniyi said.
