This is a research prototype. The data and analyses are preliminary and not yet validated — we'd welcome your .
An AI-generated image of a fake 5,000-naira banknote featuring President Bola Tinubu circulated on social media in Nigeria in January 2025. Fact-checkers confirmed the image was created using the Grok AI tool, as evidenced by a visible watermark that some users attempted to crop out. The Central Bank of Nigeria denied the existence of such a banknote, identifying the claim as misinformation.
An image circulated on Nigerian social media in late January 2025 purporting to show a newly issued 5,000-naira banknote featuring President Bola Tinubu. The Central Bank of Nigeria denied the claim, and fact-checkers reported that the image was likely generated using an AI tool linked to the platform X, based on a visible Grok watermark.
Domain classification, causal taxonomy, severity scores, and national security assessments were LLM-classified and may contain errors.
AI systems that inadvertently generate or spread incorrect or deceptive information, which can lead to inaccurate beliefs in users and undermine their autonomy. Humans that make decisions based on false beliefs can experience physical, emotional or material harms
The incident involved the generation and viral spread of a false image of a N5,000 banknote, leading to inaccurate beliefs among social media users about the Central Bank of Nigeria's policies.
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 users who intentionally generated a fake banknote image using Grok and shared it on social media to misinform the public.
Added categories peak at 2: Minor (Epistemic Harm, Direct)
An AI-generated image of a fake N5,000 Nigerian banknote featuring President Tinubu was circulated on social media. Generated using Grok, the image was quickly debunked by fact-checkers and the Central Bank of Nigeria. The incident represents a minor, human-controlled misinformation event with minor national security implications.
The report describes Grok as an 'artificial intelligence (AI) tool' used by social media users to generate a fake banknote image. It performed a specific, well-defined function (image generation) based on direct user input.
Risk Level 3. Limited Risk: The report describes the use of Grok to generate synthetic media (a fake banknote image), which falls under AI-generated content and deepfakes requiring transparency obligations.
The AI system Grok was used to generate a realistic but fake image of a banknote featuring a political figure, which was then shared on social media.
The report clearly identifies the AI tool used (Grok), the nature of the misinformation (fake N5,000 note), and the platform where it circulated (Facebook). There is little ambiguity about the events.
None
Technology
AI Image Generation Capabilities
Grok allowed generation of realistic currency notes depicting the President.
Easily Removable Watermarks
The Grok watermark was easily cropped out of the image by malicious users.
Data Inputs
Training on President Likeness
AI model trained on public figures, enabling generation of Tinubu's face.
Training on Currency Designs
AI model trained on bank note imagery, enabling realistic currency creation.
Human Factors
Intentional Deception by Users
Social media users cropped watermarks to intentionally spread misinformation.
Public Credulity
Users shared unverified fake news due to history of currency redesigns.
Process and Methods
Lack of Image Verification
Social media platforms lacked automated tools to detect cropped AI watermarks.
Lack of Prompt Filters on Grok
Grok's generation process lacked filters preventing currency counterfeits.
Management
Inadequate Platform Safeguards
X and Grok management failed to prevent misuse of image generation.
Regulatory Environment
Lack of AI Content Regulations
Absence of strict regulations in Nigeria governing AI-generated misinformation.
AI system involved
Identified by automated extraction; can differ from the alleged developers above.
