Nigeria’s Defence Headquarters has disowned an AI-generated video that falsely portrays Chief of Defence Staff General Olufemi Oluyede making partisan political statements, describing it as a deliberate attempt to create the impression that the Armed Forces is taking a political position.
In a statement issued on 27 August, Director of Defence Information Major General Samaila Uba said the video did not originate from the Chief of Defence Staff and that the statements attributed to General Oluyede were false. The material was created through AI manipulation and deliberately designed to appear authentic, the DHQ said.
The military said the video appeared intended to mislead Nigerians, distort the views of the CDS and damage the reputation of the Armed Forces, and urged the public to disregard it entirely.
Why this case differs
Nigeria has seen a run of AI-generated impersonations over the past two years, but most have been commercial. The FactCheckHub documented deepfakes of broadcaster Olamide Odekunle promoting a fraudulent trading platform and presenter Kayode Okikiolu endorsing miracle drugs, alongside deepfake advertisements using business figures including Tony Elumelu to push Ponzi schemes.
Fabricating partisan statements by the serving head of the armed forces is a different category. Nigeria’s constitutional settlement rests in part on military non-involvement in politics, and the DHQ was explicit that the danger lay in the position of the person impersonated as much as in the content.
The Armed Forces, the statement said, remains a professional, apolitical and non-partisan institution committed to the Constitution and the defence of the nation.
A prediction arriving early
The incident tracks closely with what Bloomwit Africa forecast in July, when its report on Nigeria’s 2027 vote argued the election would be fought in channels most institutions do not monitor.
The report’s central point was an asymmetry: the cost of fabricating a damaging association has collapsed, the cost of distributing it has collapsed, and the cost of detecting and correcting it has not fallen at all. With 109 million internet users and stories frequently breaking in Hausa, Yoruba or Igbo before reaching English, it argued that most organisations are “watching the wrong room” — monitoring mainstream media while narratives form in encrypted groups and forwarded voice notes.
Its advice to institutions was to assume a convincing fabrication of their spokesperson could be produced cheaply and distributed instantly, and to build detection and rebuttal capacity before the cycle peaks. The DHQ has now had to do exactly that, more than a year out from the vote.
The response, and its limits
The military’s handling was faster and more structured than most institutional responses to deepfakes on the continent. It issued a same-week denial, listed its five verified channels — Defence Headquarters Nigeria on Facebook, DHQNigeria on X, DefenceHQ on Instagram, Defence Headquarters Nigeria on WhatsApp and DefenceTV-Nigeria on YouTube — and warned those amplifying or circulating the material, not only those who created it.
What the statement does not indicate is any investigation, referral to law enforcement, or the legal route by which anyone might be pursued. Nigeria’s Artificial Intelligence Control and Regulation Bill remains before the National Assembly. Zambia, by contrast, has prosecuted AI-generated content under its existing Cyber Crimes Act, with police warning in June that offenders faced arrest ahead of that country’s elections.
Detection capacity is also thin. Research covered by iAfrica has repeatedly found African-language and African-context content poorly served by detection systems, and Meta’s phasing out of third-party fact-checking in Africa has widened the gap. A recent North-West University study extending misinformation detection into isiZulu and Sepedi found that translation quality directly determines model accuracy for low-resource languages — a finding that applies equally to Hausa, Yoruba and Igbo.
The DHQ’s advice — verify sensitive claims through official channels before believing or sharing them — is sound, and for now it is close to the only defence available.
