On May 26, 2026, a synthetic audio of Nigeria’s President went viral. An innocent man faced prosecution. And Nigeria had no tool to prove it was fake. This is not a future threat; it is today’s emergency.
Nigeria’s Presidency threatens prosecution over AI-generated audio of President Tinubu circulated on social media.
A synthetic audio clip purporting to be the voice of Nigeria’s President went viral across WhatsApp, X, and Instagram. Millions heard it. Many believed it. Within hours, the Presidency called for the prosecution of a social media personality accused of spreading it. By nightfall, investigations showed he may not have even posted it. An unknown actor had taken his real video, layered a fake voice over it, and released it into Nigeria’s information ecosystem and walked away.
Nobody was ready. Not the Presidency. Not the platforms. Not the public. Not the law.
I have been saying this for a while, most recently at a stakeholders workshop in Lagos, just hours before this incident broke. I said that Nigeria’s AI harm reality is fundamentally different from what the global north talks about when they talk about AI safety. I said that deepfakes and synthetic media would be our greatest harm. I said that if we do not build governance infrastructure now, we will be managing crises instead of preventing them.
I did not expect to be proven right the same day.
“An unknown actor layered a fake voice over a real video, released it into Nigeria’s information ecosystem and walked away. Nobody was ready. Not the Presidency. Not the platforms. Not the public. Not the law.”
Dr. Ayuba Tuodolo, AI Safety Nigeria
What Exactly Happened
On May 26, 2026, an audio clip purporting to feature President Bola Ahmed Tinubu went viral across Nigerian social media. The voice was not his. It was generated by artificial intelligence and layered over an authentic Instagram video posted by a social media activist. Martins Vincent Otse, known as VDM.
Presidential spokesperson Bayo Onanuga reacted swiftly, calling for VDM to “face the full weight of the law” for spreading a fake audio. The problem? Further investigation established that VDM had not uploaded the fabricated audio himself. An unidentified actor had repurposed his footage, added synthetic audio, and circulated it across platforms, and the Presidency had aimed its legal threats at the wrong person.
This is what makes this case so dangerous: the Presidency could not verify whether the audio was real or fake. A Nigerian citizen nearly faced criminal prosecution for content he did not create. The real perpetrator is still unknown. And this could happen to any one of us.
This is not the first time. In 2024, a deepfake video of the same President went viral, fabricated footage of him claiming he wanted to buy Chelsea Football Club. Before that, the 2023 elections were flooded with AI-generated audio of opposition candidates allegedly plotting to rig the election. In each case, the harm spread faster than the truth. In each case, Nigeria had no system to respond.
Nigeria’s Growing Deepfake Timeline:
- 2023 elections: AI-generated audio of presidential candidates “plotting” election rigging circulated hours before polling
- 2023: Deepfake videos of Hollywood actors and Elon Musk “endorsing” presidential candidates spread across social media
- 2024: AI-generated video of President Tinubu claiming to want to buy Chelsea FC.
- 2024: Fake investment advertisements using manipulated visuals of the President promoted suspected Ponzi schemes
- 2025: An AI-generated video of Donald Trump falsely calling for the release of Nnamdi Kanu circulated on Facebook
- 26 May 2026: Synthetic audio of President Tinubu goes viral – innocent man faces prosecution, real creator unknown
Our Reality Is Not the Global Conversation
When the most powerful AI companies talk about AI harm, they talk about existential risk, autonomous weapons, and the alignment of superintelligent systems. These are legitimate concerns. But they are not Nigeria’s most urgent concerns today.
Nigeria’s AI harm landscape is dominated by synthetic media weaponised for political manipulation, financial fraud, and social destabilisation. It is dominated by AI-generated content in Yoruba, Hausa, Igbo, and Pidgin, languages that global content moderation systems barely recognise, let alone monitor. It is dominated by platforms that have no local language classifiers and no cultural context for what constitutes dangerous speech in the Nigerian information environment.
When AI-generated investment scam videos impersonating local pastors and celebrities circulate in Yoruba, Meta’s content moderation does not catch them because it was not built for Yoruba. When a synthetic audio in Hausa is used to incite communal violence, there is no incident reporting system in Nigeria to receive that report. When a deepfake is used to frame a political opponent ahead of a 2027 election, there is no forensic AI audit authority to investigate it.
The global AI safety conversation is not designed to solve these problems. We must solve them ourselves.
“Nigeria’s AI harm landscape is dominated by synthetic media in Yoruba, Hausa, Igbo, and Pidgin languages that global content moderation systems barely recognise, let alone monitor.”
— Dr. Ayuba Tuodolo, AI Safety Nigeria
What the Law Says and Where It Falls Short
Nigeria is not without legal tools. The Cybercrime Act (amended 2024) criminalises the creation and distribution of false digital content intended to damage reputations or mislead the public. The E-Governance Bill 2025, currently before the National Assembly, goes further, and it is worth examining what it actually says.
What the E-Governance Bill 2025 Provides (Parts XI–XII)
AI systems must be transparent and explainable, and users must understand how decisions are made.
Citizens have the right to challenge AI decisions that affect them and seek redress
Developers must adopt safeguards to detect discriminatory outcomes and protect individuals from AI-generated harm
The Regulatory Agency must accredit independent AI auditors, including for deepfake verification.
The Agency must promote public education on AI risks and support capacity building
An annual AI impact assessment report must be published by the Minister
Maximum penalty for violations: ₦10,000,000 ($6,250) or 2% of Nigerian revenue
These are meaningful provisions.
The Bill is a significant progress. But here is the critical problem: the Presidency itself did not have access to an AI forensic tool to verify whether the audio was real or synthetic. The right to challenge AI-generated content means nothing if you cannot prove it was generated by AI in the first place.
A law without an enforcement infrastructure is decoration. The E-Governance Bill creates rights, but it does not yet create the institutions, tools, or technical capacity to make those rights real. That gap is where people get hurt. That gap is where innocent people face prosecution for content they did not create.
Three Things That Must Happen Quickly
I am not writing this to criticise the government. I am writing this because I have spent years building the case that Nigeria needs a different kind of AI governance, one grounded in Nigerian realities, not borrowed frameworks. This incident is the most visible proof yet of why that work matters.
First, Nigeria needs a national AI forensic verification capability. The Presidency should not have to rely on social media users and independent journalists to determine whether an audio is real. NITDA or a designated AI Safety Directorate must commission and deploy open-source AI detection tools, accessible to government agencies, media organisations, and citizens, that can verify whether audio, video, and images are AI-generated. This technology exists. Nigeria does not need to build it from scratch. It needs to deploy and localise it.
Related:
Second, Nigeria needs a national AI harm registry. Every incident of AI-generated harm, deepfakes, synthetic scam content, and biased automated decisions must be documented in a publicly accessible, community-sourced database. AI Safety Nigeria has proposed building this registry. It should be seeded by civil society, hosted by NITDA, and updated continuously. Without a harm record, we cannot demonstrate the scale of the problem, we cannot hold platforms accountable, and we cannot measure whether our laws are working.
Third, the E-Governance Bill must pass with the Regulatory Agency clearly defined. The Bill’s most dangerous weakness is that it creates obligations without clearly naming which institution will enforce them. Every day the Bill sits in the National Assembly undefined, is another day Nigerians have rights on paper with no mechanism to exercise them.
“A law without enforcement infrastructure is decoration. The E-Governance Bill creates rights – but it does not yet create the institutions, tools, or technical capacity to make those rights real.”
— Dr. Ayuba Tuodolo, AI Safety Nigeria
If the Presidency Is Not Safe, Neither Are You
Here is what I want every Nigerian to understand from the incident: if AI can be used to fabricate the voice of the most powerful person in this country convincingly enough to fool millions and nearly destroy an innocent man, it can be used to fabricate yours.
Your voice. Your image. Your reputation. None of them is safe in an environment without an AI governance infrastructure. A landlord can use a deepfake to frame a tenant. A business rival can use synthetic audio to destroy a competitor. A political opponent can use an AI-generated video to end a career. And in every case, Nigeria currently has no system to report it, investigate it, attribute it, or remedy it.
This is not a technology problem. It is a governance problem. And governance problems have governance solutions if the political will exists to pursue them.
AI Safety Nigeria exists because someone has to build the pressure for that political will. Our fellowship of 91 practitioners across Nigeria and Africa, our policy research, our community advocacy, all of it is oriented toward the day when Nigeria has the governance infrastructure to protect its citizens from exactly what happened yesterday.
That day is not here yet. But it must come before 2027.
Dr. Ayuba Tuodolo
He is the Founder and Executive Director of AI Safety Nigeria, driving Nigeria’s most comprehensive AI literacy and safety movement. He is the convener of the first AI Safety Summit in Nigeria and the African Cybersecurity Convention, serves on the Advisory Council of Blacks in AI Safety and Ethics (BASE) USA, and brings over 15 years of cybersecurity expertise to the AI governance space.