This is a research prototype. The data and analyses are preliminary and not yet validated — we'd welcome your .
Scammers are using AI-generated images and videos to impersonate celebrities, including Brad Pitt and Jennifer Aniston, to conduct long-term romance scams. Victims are manipulated through emotional bonding and fake crises, such as medical emergencies, to transfer large sums of money. These incidents have resulted in significant financial loss, severe mental health impacts on victims, and subsequent online harassment.
Three Nigerian "Yahoo Boys" scammers allegedly used AI-generated image manipulation tools, along with fake social media and WhatsApp accounts, to reportedly impersonate actor Brad Pitt and convince a French interior decorator, Anne, that she was in a romantic relationship with him. Over 18 months, they allegedly fabricated selfies and messages, reportedly leading Anne to divorce her husband and transfer $850,000 under the false pretense that Pitt needed money for kidney treatment while his accounts were frozen.
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
Vulnerable individuals placed inappropriate trust in AI-generated celebrity personas, leading to severe emotional and material dependence.
The scammers used AI-generated deepfakes to impersonate celebrities and execute targeted romance scams, defrauding victims of significant financial sums.
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 scammers who intentionally used deployed AI image and video generation tools to defraud specific victims.
Added categories peak at 4: Severe (Epistemic Harm, Inferred)
Scammers used AI-generated deepfake images and videos to impersonate celebrities in sophisticated romance scams, defrauding individuals of significant sums and causing severe psychological distress. While devastating to the individuals, the national security impact remains minor and manageable under standard law enforcement procedures.
The AI systems (image and video generators) were used as tools by human scammers to generate specific synthetic media assets based on direct inputs. The AI did not exhibit autonomous planning or independent decision-making.
Limited Risk: The report describes the use of AI-generated content such as deepfakes, which are subject to transparency obligations under the EU AI Act to ensure users are informed of their synthetic nature.
The primary function of the AI systems in this incident was to generate highly realistic synthetic images and videos of celebrities to deceive the victims.
The reports provide consistent, detailed accounts of the two primary scam occurrences, including specific financial losses, the psychological impact on the victims, and the specific ways AI-generated images and videos were used to facilitate the fraud. The role of the AI as a tool for generating deceptive media is clearly outlined across multiple reputable news sources.
The incident could be viewed purely as a traditional cybercrime/phishing incident, but the central role of AI-generated deepfakes in overcoming the victims' skepticism makes it a clear AI safety/misuse incident.
Technology
Convincing Deepfake Generation
AI tools created realistic hospital photos and video clips of celebrities.
Accessible Voice and Video AI
Scammers easily cloned celebrity voices and created fake news reports.
High-Fidelity Image Editing
AI allowed seamless face swapping and custom text insertion on images.
Data Inputs
Exposed Celebrity Personas
Abundant public data allowed scammers to train and craft convincing deepfakes.
Public Social Media Posts
Scammers used public ski trip photos to target victims.
Doctored Digital ID Inputs
Scammers used modified digital copies of driving licenses to prove identity.
Human Factors
Low Social Media Literacy
Victims did not understand platform mechanics or how to spot deepfakes.
Emotional Vulnerability
Scammers exploited victims struggling with divorce, isolation, or depression.
Overlooking Red Flags
Victims ignored warnings from family and signs like mismatched lighting.
Process and Methods
Weak Platform Verification
Social media sites allowed unverified accounts to impersonate celebrities.
Delayed Content Moderation
Platforms failed to proactively flag or remove deepfake scam accounts.
Inadequate Reporting Systems
User reporting mechanisms did not stop scammers from continuing harassment.
Management
Prioritizing Growth Over Safety
Platforms delayed implementing automatic scam flagging to maintain user growth.
Poor AI Tool Risk Assessment
Developers released generative models without safeguards against fraud.
Regulatory Environment
Lack of AI Legislation
Governments failed to establish laws protecting citizens from deepfake abuse.
Weak Cross-Border Enforcement
Jurisdictional issues hindered prosecuting scammers operating from Nigeria.
AI system involved
Identified by automated extraction; can differ from the alleged developers above.
