2027: Experts outline ways to detect AI-manipulated content

As Nigeria prepares for the 2027 general elections, journalists have been urged to strengthen their verification processes as part of efforts to combat artificial intelligence-manipulated content and misinformation.

Media stakeholders also urged journalists and newsrooms to scrutinise suspicious digital content and establish clear newsroom policies on the use of AI to tackle the growing threat of misinformation.

They made the call on Wednesday during a PUNCH Media Foundation webinar on artificial intelligence, misinformation and Nigeria’s 2027 general elections.

Speaking during the webinar, the Founder of FactsMatterNG, Hannah Ajakaiye, said journalists must pay close attention to the timing, origin and quality of information, particularly audio recordings, videos and images circulated shortly before elections.

Offering insights into frontline verification, synthetic content and information integrity, Ajakaiye said the timing of a suspicious publication could provide an important clue to its credibility.

She said, “The timing of the content is critical. If an election is imminent and on the eve of the election, someone pushes out an audio clip, video, or manipulated image, the timing is very important, and that should raise red flags.

“You must be sure of the origin. You cannot simply rely on something forwarded on WhatsApp; you have to trace it to its primary source.”

Ajakaiye also urged journalists to contact relevant institutions directly to authenticate claims rather than rushing to publish unverified material in a bid to break the news.

She said, “Before pushing content out in a rush to break the news, journalists must reach out to the relevant institutions to verify. In this context, you reach out directly to INEC. In my organization, there have been times when we had to step down a story to verify it first, rather than risking our reputation.”

She said journalists should also be cautious of messages designed to create panic or pressure recipients into sharing information immediately, noting that such tactics could be used to spread manipulated content.

“Be wary of content that pressures you to act immediately—messages containing phrases like ‘Share this right now!’ or ‘Share before it gets deleted!’ For non-journalists, if you see a prompt urging you to urgently click or share a specific link, that sense of urgency and emotional triggers are something to watch out for,” she said.

The FactsMatterNG founder further recommended checking whether credible media organisations or official institutions had already debunked a claim before treating it as factual.

“Check whether credible media platforms or official institutions have already debunked the claim. In many instances, deepfakes production are low quality. If it is audio, listen closely for unnatural pauses, muffled tones, or inconsistent tracks, which indicate that different audio clips were spliced together. Low-quality audio, or content published without a clear, traceable source, are all signs to stop and verify before publishing,” she said.

Similarly, the Director of Projects at the Cable Newspaper Journalism Foundation, Mayowa Tijani, warned that the 2027 elections could witness more sophisticated attempts to manipulate information than previous electoral cycles.

Tijani, who spoke on AI, election reporting, newsroom standards and editorial trust during the webinar, said the widespread adoption of AI had increased the potential for manipulated content.

He said, “One of the things that is obvious to me and has been revealing itself a lot more as we got closer to the elections is that we are going to see a lot more manipulations than we have seen before.

“During the 2023 elections, it was just a year AI went mainstream, and now we have three years in which AI has been widely adopted.”

He called on media organisations to develop and publicly disclose clear policies explaining how they use AI in their newsrooms, arguing that such guidelines would help audiences understand the role of the technology in news production.

“One of the things that we need to do, and I can recommend this very strongly, and this is a conversation that has been going on among newsroom leaders in the last couple of months, is that the majority of the newsrooms in Nigeria, possibly every newsroom in Nigeria, should have a public task-specific disclosure of AI policy, that clearly explains to their audience what they use AI for, where they use AI,” he said.

Tijani said the absence of clear policies had created inconsistencies in how journalists and editors approached the use of AI, sometimes leaving different members of the same newsroom with conflicting standards.

He said, “One of the primary issues we face is the absence of policy. Different newsrooms are held to different standards, and even within the same newsroom, some people are singing under different hymn sheets. So we have a situation in the newsroom where a certain person believes it’s fine to use AI to write news.

“Even within a newsroom, different people are singing from different hymn sheets, and so we have a situation where a certain person in the newsroom believes that it’s fine to use AI to write the story, and another person in that same newsroom believes that AI should not be used in this particular instance.”

He noted that audiences were not necessarily opposed to media organisations using AI but wanted transparency about the purposes for which the technology was deployed and how it influenced published reports.

“What they basically want to know is what you’re using it for. They want that specific disclosure on the use of AI, because one of the things that happens when we don’t do that is that we ourselves can then be the source of misinformation,” he said.

Tijani urged media organisations to publish their AI policies well ahead of the elections to give newsrooms and audiences sufficient time to understand the guidelines before the heightened political activities associated with the polls.

He also distinguished between tasks that AI could support and responsibilities that required human editorial judgement, stressing that the final decision on what gets published must remain with journalists and editors.

“So I think that much of our writing work, transcription, and a lot of the work that journalists do basically can be well done by AI. Now, the part of our work that is not easily done by AI is the part of judgement, human judgement, and I feel like that’s where like much of the editorial burden should be at this time.

“So as long as there is the human in the loop who has editorial judgement, I think that a lot of the risk that we naturally face with AI use is sort of removed,” Tijani said.