This is a research prototype. The data and analyses are preliminary and not yet validated — we'd welcome your .
A viral deepfake video on Facebook falsely depicts Kenyan journalist Jeff Koinange endorsing a mobile gambling app. The video uses manipulated footage from a legitimate Citizen TV broadcast, altering the journalist's voice and lip movements. Africa Check confirmed the video is fraudulent, noting inconsistencies in the journalist's accent and the repurposing of unrelated footage.
A viral video purportedly manipulated by AI shows Kenyan journalist Jeff Koinange endorsing a mobile gambling app. The video was reportedly created using AI deepfake techniques, uses footage from a legitimate news broadcast, and artificially alters Koinange's voice and mouth movements. Neither Koinange nor Citizen TV has promoted the app, and no legitimate source hosts the video.
Domain classification, causal taxonomy, severity scores, and national security assessments were LLM-classified and may contain errors.
Using AI systems to gain a personal advantage over others such as through cheating, fraud, scams, blackmail or targeted manipulation of beliefs or behavior. Examples include AI-facilitated plagiarism for research or education, impersonating a trusted or fake individual for illegitimate financial benefit, or creating humiliating or sexual imagery.
Also classified as
The deepfake video generates and spreads false information by depicting a well-known journalist endorsing a product he did not actually support.
The incident involves the use of AI-generated deepfakes to impersonate a trusted public figure to promote a fraudulent gambling app for personal gain.
Human
Due to a decision or action made by humans
Intentional
Due to an expected outcome from pursuing a goal
Post-deployment
Occurring after the AI model has been trained and deployed
The incident was caused by human actors who intentionally used deployed AI tools to create a deceptive deepfake video for financial gain.
Added categories peak at 2: Minor (Epistemic Harm, Direct)
An AI-generated deepfake video impersonated prominent Kenyan journalist Jeff Koinange to promote a fraudulent gambling app, reaching over 227,000 views on Facebook. While representing a minor threat to informational and societal trust, the incident is a localized financial scam with negligible direct implications for national security, critical infrastructure, or state sovereignty.
The AI was used as a tool by human creators to analyze voice and facial movements and generate a deepfake video based on direct inputs.
Limited Risk: The incident involves 'AI-generated content such as deepfakes' which requires transparency obligations under the EU AI Act.
The AI system was used to generate a deepfake video and clone the voice of the journalist to make it appear as though he endorsed the app.
The report from Africa Check is highly detailed, providing clear evidence of the deepfake manipulation, the original source video, and the platform where it was distributed. The role of AI in generating the fake endorsement is explicitly verified.
None. The evidence clearly points to an intentional deepfake scam.
Technology
AI Voice Cloning Technology
AI cloned the journalist's voice to make the scam sound authentic.
Realistic Facial Synthesis
AI tools matched lip movements to the fake audio convincingly.
Data Inputs
Unrestricted Source Footage
Public YouTube videos of the journalist were used to train the AI.
Human Factors
Trust in Known Journalists
Viewers trust recognizable presenters, making them vulnerable to scams.
Process and Methods
Weak Platform Moderation
Facebook's automated filters failed to flag the deepfake prior to upload.
Lack of Content Verification
No process verified the authenticity of the video before distribution.
Management
Prioritizing Reach Over Safety
Platforms focused on content distribution rather than verification.
Regulatory Environment
Weak AI Impersonation Laws
Lack of regulations governing deepfakes and AI-generated content.
AI system involved
An AI system was involved, but the reports do not identify it.
Identified by automated extraction; can differ from the alleged developers above.
