Purported Unauthorized Deepfakes of Norman Swan and Others Circulated in Online Supplement Campaigns

According to an ABC News (Australia) 7.30 report, a second wave of deepfake scam ads impersonating Norman Swan and other public figures circulated widely on Meta platforms in 2025, promoting unproven health supplements. The campaign featured new voice-cloned videos and more impersonation tactics, and it reportedly contributed to financial and health harms. Investigators reportedly linked the ads to businesses in Australia and New Zealand using deceptive redirects and alias branding.

Risk classification

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

The incident involves scammers using AI-generated deepfakes of trusted figures to deceive specific consumers for financial gain and unauthorized credit card charges.

Additional risk subdomains

• 3.1 False or misleading information: The deepfakes spread false health advice, leading a diabetes patient to stop taking critical prescribed medication.

Causal factors

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

The deepfake videos were intentionally created and deployed by human scammers to defraud consumers and promote unproven products.

EU AI Act risk tier

• Risk tier: 3 Limited Risk

Limited Risk: The report describes the use of AI-generated content such as deepfakes, which falls under Risk Level 3 and is subject to transparency obligations under the EU AI Act.

AI system and alleged parties

• AI system: unspecified
• AI purpose: Deepfake Video Generation; Voice Generation
• Behaviour type: Tool
• Alleged developer: Unknown voice cloning technology developers, Unknown deepfake technology developers
• Alleged deployer: Unidentified actors linked to Vellec Group NZ Ltd, Unidentified actors linked to Healthy Life Choices NZ Ltd, Unidentified actors linked to Apex United Pty Ltd
• Alleged harmed parties: Rebel Wilson, Public trust in medical guidance, Norman Swan, Jonathan Shaw, General public of Australia, Australians with chronic health problems, Adele

Harm severity

Severity ratings are based on this scale.

Highest direct severity in any category: Minor. 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 Minor
• Infrastructure: direct Negligible, indirect Negligible
• Property: direct Negligible, indirect Negligible
• Financial: direct Negligible, indirect Negligible
• Environmental: direct Negligible, indirect Negligible
• Malicious content: direct Minor, indirect Minor
• Differential treatment: direct Negligible, indirect Negligible
• Civil rights: direct Negligible, indirect Negligible
• Democracy: direct Negligible, indirect Negligible
• Privacy: direct Negligible, indirect Negligible
• Psychological: direct Negligible, indirect Minor
• Epistemic: direct Minor, indirect Substantial
• Child sexual exploitation and abuse: direct Negligible, indirect Negligible

Physical

Reported: The report explicitly describes physical health risks and harm, including a patient stopping critical diabetes medication and another gaining weight.

Indirectly caused: A diabetes patient stopped taking metformin, a core drug preventing diabetes complications, due to the deceptive deepfake. Another victim gained weight while taking the unproven supplement.

Inferred additional harm: It is highly likely that other patients among the thousands exposed also stopped critical medications, leading to unquantified but severe adverse health outcomes.

Malicious content

Reported: The report explicitly describes the creation and spread of malicious deepfake videos used to scam people.

Directly caused: AI deepfake software was used to generate deceptive videos of doctors and celebrities to promote unproven health supplements.

Indirectly caused: These malicious videos were widely distributed on Meta platforms, reaching thousands of users.

Inferred additional harm: Thousands of similar malicious deepfake ads are likely generated and spread across social media platforms daily.

Psychological

Reported: The report explicitly describes emotional distress experienced by a victim of the scam.

Indirectly caused: One victim reported feeling 'shattered' and describing the experience of being defrauded as 'horrible'.

Inferred additional harm: Many of the numerous other defrauded customers likely experienced similar distress, anxiety, and anger upon realizing they were scammed.

Epistemic

Reported: The report explicitly describes epistemic harm, noting that these deepfakes cause public mistrust in legitimate medical advice.

Directly caused: The AI deepfakes fabricated statements from medical professionals denigrating science and promoting unproven treatments.

Indirectly caused: This leads to a broader erosion of trust in public health authorities and medical advice, as noted by Professor Shaw.

Inferred additional harm: Widespread proliferation of medical deepfakes could severely degrade the public's ability to distinguish real medical advice from scams, leading to systemic public health crises.

People affected

• Occurrences reported: 1
• People reportedly harmed: 10
• People reportedly exposed: 10000

Potential causes

Management

• Prioritizing Ad Revenue: Meta continues to accept ad revenue from scam accounts running deepfakes.
• Inadequate Risk Assessment: Failure to evaluate public safety risks before changing moderation policies.

Technology

• Accessible Deepfake Software: Low-cost software allows easy cloning of voices and faces of trusted figures.
• Sophisticated AI Generation: AI models map faces and clone voices from short video samples accurately.
• Exploitation of Ad Algorithms: Scammers leverage Meta's targeting algorithms to reach vulnerable users.

Data Inputs

• Public Source Media Use: Scammers easily scrape public YouTube videos to train deepfake models.
• Lack of Media Verification: Meta's ad system lacks verification of the media assets uploaded for ads.

Human Factors

• Trust in Authority Figures: Victims trust recognized medical experts, making them vulnerable to scams.
• Vulnerability of Patients: People with chronic health issues are easily targeted by weight loss ads.
• Cognitive Bias Toward Realism: Highly convincing deepfakes make it difficult for users to detect fraud.

Process and Methods

• Weakened Content Moderation: Meta's January 2025 policy change led to increased harmful content.
• Ineffective Takedown Process: Meta's slow response to user reports allows scam ads to remain active.
• Inadequate Ad Screening: Meta fails to proactively filter deepfake ads before they go live.

Regulatory Environment

• Delayed Regulatory Penalties: Fines for platforms failing to remove scams are not yet fully enforced.
• Platform Liability Loophole: Platforms claim they are not publishers to avoid liability for scam ads.

Information quality

• Classification confidence: High
• Reason for confidence: The reports provide detailed, first-hand accounts from both the impersonated medical professionals and the victims who fell for the scams. The technical process of creating the deepfakes is explained, and the financial and health impacts are well-documented.
• Ambiguities identified: The exact deepfake software used is not named, and the precise number of total victims is not quantified beyond 'numerous'.
• Alternative interpretations: None. The incident is clearly a coordinated deepfake-enabled financial and health scam.

A widespread deepfake scam campaign on Meta platforms targeted Australian consumers, impersonating medical professionals to sell unproven supplements. While causing individual health risks and financial losses, the incident represents a low-level consumer fraud threat rather than a direct challenge to national security.

• Overall national security impact: Minor
• Scope: Single nation
• Primary target: Australia
• Alleged perpetrator: Unknown

Threat characteristics

• Imminence: Long-term. Represents an ongoing strategic and cybercrime threat vector rather than an active, immediate national security crisis.
• Autonomy: Human-controlled. AI software acts as a tool to clone voices and map faces based on direct human input, without autonomous execution or decision-making.
• Novelty: Established threat. Deepfake-enabled consumer scams and financial fraud are established threats that have occurred in various forms previously.

Impact by dimension

• Physical security: Negligible. No physical kinetic attacks or critical infrastructure disruptions reported; health risks are limited to individual consumer decisions based on misleading ads.
• Information security: Minor. Involves deepfake technology to generate misleading content, but targeting is limited to commercial scams rather than state-sponsored information warfare or institutional disruption.
• Sovereignty: Negligible. No evidence of threats to state authority, electoral processes, or core government administrative decision-making.
• Economic security: Minor. Causes financial losses to individual citizens via unauthorized recurring charges, but does not pose a systemic threat to national financial systems or strategic technologies.
• Societal stability: Minor. Causes localized epistemic harm and erodes trust in medical professionals among victims, but does not trigger large-scale civil unrest or systemic human rights violations.