Purportedly AI-Generated Deepfake Ads on Instagram Impersonate Gisele Bündchen and Other Celebrities in Brazilian Fraud Scheme

Police in Brazil reportedly arrested four suspects accused of using purportedly AI-generated deepfake videos of model Gisele Bündchen and other celebrities in Instagram ads to promote fake giveaways and skincare products. The scheme was allegedly active since at least August 2024 and generated over 20 million reais ($3.9 million) in suspicious funds. Victims allegedly lost small amounts that often went unreported, creating what investigators called "statistical immunity."

Risk classification

• Primary risk domain: 4 Malicious actors
• Primary risk subdomain: 4.3 Fraud, scams, and targeted manipulation

The incident involves a criminal group using AI-generated deepfakes of celebrities to conduct targeted financial fraud and scams at scale.

Additional risk subdomains

• 3.1 False or misleading information: The scammers generated false video advertisements depicting celebrities promoting non-existent products to mislead consumers.

Causal factors

• Entity: Human
• Intent: Intentional
• Timing: Post-deployment

The fraud was orchestrated by human criminal actors who intentionally utilized deployed AI deepfake technology to deceive victims for financial gain.

EU AI Act risk tier

• Risk tier: 3 Limited Risk

Limited Risk: The AI system was used to generate deepfakes. According to the EU AI Act definitions, 'AI-generated content such as deepfakes' falls under Risk Level 3 (Limited Risk), which requires transparency obligations to inform users.

AI system and alleged parties

• AI system: unspecified
• AI purpose: Deepfake Video Generation; Social Media Content Generation
• Behaviour type: Tool
• Alleged developer: Unknown voice cloning technology developers, Unknown deepfake technology developers, Meta
• Alleged deployer: Unknown scammers operating Instagram ad accounts, Unidentified Brazilian fraud group (arrested in Rio Grande do Sul)
• Alleged harmed parties: Sabrina Sato, Maisa Silva, Juliette Freire Feitosa, Gisele Bündchen, General public of Brazil, Brazilian consumers defrauded via Instagram ads, Angélica Huck

Harm severity

Severity ratings are based on this scale.

Highest direct severity in any category: Substantial. Severity is scored from Negligible to Catastrophic in each harm category, for harm the reports describe as caused directly or indirectly by the AI system.

• Physical: direct Negligible, indirect Negligible
• Infrastructure: direct Negligible, indirect Negligible
• Property: direct Negligible, indirect Negligible
• Financial: direct Substantial, indirect Negligible
• Environmental: direct Negligible, indirect Negligible
• Malicious content: direct Minor, indirect Negligible
• Differential treatment: direct Negligible, indirect Negligible
• Civil rights: direct Negligible, indirect Negligible
• Democracy: direct Negligible, indirect Negligible
• Privacy: direct Minor, indirect Negligible
• Psychological: direct Negligible, indirect Negligible
• Epistemic: direct Minor, indirect Negligible
• Child sexual exploitation and abuse: direct Negligible, indirect Negligible

Financial

Reported: Yes, the reports state that the fraudulent scheme moved over 20 million reais (approximately 3.9 million US dollars) in suspicious funds, with individual victims typically losing under 100 reais ($19) each.

Directly caused: Victims lost money paying for shipping fees of non-existent products, with individual losses typically under 100 reais ($19) per occurrence, contributing to a total of over 20 million reais ($3.9 million) in suspicious transactions.

Inferred additional harm: The total financial loss could be higher as many victims did not report their losses due to the small individual amounts, and the justice system froze up to 210 million reais in assets associated with the group.

Malicious content

Reported: Yes, the reports describe the creation and dissemination of deepfake videos of celebrities promoting fake products.

Directly caused: The criminal group created deepfake videos of Gisele Bundchen and other celebrities to promote non-existent products on social media platforms like Instagram and Facebook.

Inferred additional harm: It is likely that many other deepfake videos and fraudulent ads were generated and distributed across social media platforms that were not individually detailed in the reports.

Privacy

Reported: Yes, the reports describe the unauthorized use of celebrities' likenesses and images to create deepfakes.

Directly caused: The likenesses, images, and voices of Gisele Bundchen, Neymar, Juliette, Maisa, and Sabrina Sato were misappropriated without their consent to generate deepfake promotional videos.

Epistemic

Reported: Yes, the reports describe the use of deepfakes to fabricate celebrity endorsements for non-existent products, misleading consumers.

Directly caused: The scammers generated realistic deepfake videos of Gisele Bundchen and other celebrities to falsely show them recommending products like a 'free anti-wrinkle kit'.

Inferred additional harm: The widespread use of celebrity deepfakes in advertising contributes to a broader erosion of trust in digital media and online advertisements.

People affected

• Occurrences reported: 1
• People reportedly harmed: 6
• People reportedly exposed: 110000

Potential causes

Management

• Ad Revenue Prioritization: Platforms prioritized ad distribution over rigorous safety verification.
• Inadequate Risk Assessment: Lack of proactive risk mitigation for AI-enabled ad fraud.

Technology

• AI Deepfake Video Generation: AI tools were used to create realistic fake videos of celebrities.
• Inadequate Automated Detection: Social media platforms failed to automatically detect deepfake ads.

Data Inputs

• Unauthorized Celebrity Likeness: Using public images and voice data of celebrities to feed AI models.

Human Factors

• Victim Cognitive Bias: Victims trusted deepfakes because they featured famous celebrities.
• Malicious AI Exploitation: Fraudsters actively used AI tools to scale their deceptive schemes.

Process and Methods

• Ineffective Ad Review Process: Meta processes allowed fraudulent deepfake ads to be approved.
• Lack of Swift Content Takedown: Slow response in removing reported scam ads allowed them to run.

Regulatory Environment

• Lagging Platform Liability Laws: Platform liability for user ads was only recently established.
• Statistical Immunity Exploitation: Scammers exploited low reporting rates for small-value frauds.

Information quality

• Classification confidence: High
• Reason for confidence: The reports from CNN Brasil and Reuters provide consistent, detailed accounts of the police operation, the mechanics of the fraud, the financial scale (20 million reais), and the specific celebrities targeted. The role of AI (deepfakes) is explicitly stated and central to the scam.
• Ambiguities identified: The specific AI models or software tools used to generate the deepfakes are not identified in the reports.
• Alternative interpretations: None. The incident is clearly a case of AI-enabled financial fraud.

Brazilian police dismantled a criminal network using celebrity deepfakes for large-scale financial fraud, generating 20 million reais. While demonstrating an evolved capability in organized cybercrime and synthetic media training, the incident remains a law enforcement issue with minor national security implications.

• Overall national security impact: Minor
• Scope: Single nation
• Primary target: Brazil
• Alleged perpetrator: Brazilian criminal organization led by Cleiton da Silva Luz

Threat characteristics

• Imminence: Long-term. The specific operation was dismantled by police, representing a long-term strategic concern regarding the proliferation of deepfake fraud techniques.
• Autonomy: Human-controlled. AI was utilized strictly as a tool by human actors to generate synthetic media, with no autonomous decision-making.
• Novelty: Evolved capability. While deepfake fraud is established, the scale of the operation and the formalization of a 'digital crime university' represent an evolved threat model.

Impact by dimension

• Physical security: Negligible. No physical systems, critical infrastructure, or human safety threats were involved in this digital fraud incident.
• Information security: Minor. Involves sophisticated celebrity deepfakes, but lacks state-actor involvement, intelligence compromise, or targeting of political leadership.
• Sovereignty: Negligible. No impact on state authority, border control, or core government decision-making processes.
• Economic security: Minor. Financial fraud totaling 20 million reais represents a significant criminal enterprise but does not threaten national economic stability.
• Societal stability: Minor. Widespread consumer fraud eroding public trust in digital media, though manageable within standard law enforcement procedures.