This is a research prototype. The data and analyses are preliminary and not yet validated — we'd welcome your .
A deepfake video of entrepreneur Steven Bartlett was circulated on Instagram in April 2025 to promote a fraudulent WhatsApp investment group. The video used AI-generated content to impersonate the investor and entice users with promises of rapid stock market returns. The victim's team confirmed the video was fake and worked with platforms to remove it, highlighting the ongoing risk of AI-enabled investment scams.
A purported deepfake video of entrepreneur and BBC Dragon's Den investor Steven Bartlett circulated on Instagram in April 2025, allegedly claiming to offer exclusive stock tips and inviting users to join a WhatsApp trading group. The reported video referenced U.S. tariff policies to entice viewers with promises of rapid returns. Bartlett's team confirmed the video was fake and worked with platforms to remove it, warning that scammers were attempting to harvest contact details for further fraud.
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.
The incident involves the targeted use of a synthetic celebrity deepfake to impersonate a trusted individual for the purpose of executing an investment scam.
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 created and deployed a deepfake video of Steven Bartlett to defraud social media users.
Added categories peak at 2: Minor (Epistemic Harm, Direct)
A generative AI deepfake of entrepreneur Steven Bartlett was used on Instagram to promote a fraudulent investment group. The incident represents a common cybercrime and fraud issue rather than a national security threat, with minor implications for economic security and information integrity.
The AI was used as a tool by human scammers to generate a synthetic video (deepfake) of Steven Bartlett. The report does not describe the AI system exhibiting autonomous decision-making or agentic planning.
Limited Risk: The system is used to generate AI-generated content such as deepfakes, which under the EU AI Act are subject to transparency obligations to inform users that the content is synthetic.
The AI was used to generate a synthetic video and voice of Steven Bartlett to make it appear as though he was endorsing the investment scheme.
The report clearly describes the nature of the incident (a deepfake scam on Instagram impersonating Steven Bartlett) and the mechanisms of the fraud. There is little ambiguity about the AI's role as a generative tool used for impersonation.
None. It is clearly a malicious impersonation scam.
Technology
Advanced Deepfake Technology
AI tools generate highly realistic video and audio of public figures.
Inadequate Detection Tech
Platform detection systems fail to catch all sophisticated deepfakes.
Data Inputs
Unauthorized Celebrity Media
Scammers use public video and audio of celebrities to train AI models.
Human Factors
Cognitive Biases of Users
Scammers exploit user greed and fear of falling markets to lure victims.
Trust in Celebrity Endorsements
Users trust familiar public figures, reducing their skepticism of the scam.
Process and Methods
Inadequate Content Moderation
Platform review processes fail to block fake ads prior to publication.
Slow Takedown Procedures
Delayed removal of reported scam accounts allows the fraud to spread.
Management
Insufficient Platform Action
Meta's efforts to crack down on celebrity scams are currently incomplete.
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.
