Celebrities Targeted by Deepfake Fraud

The misuse of AI deepfake systems for impersonating public figures and ordinary individuals has directly led to financial and personal harms (fraud, privacy and portrait rights violations, reputational damage). These are realized harms caused by AI system use, fitting the definition of an AI Incident.

This describes actual misuse of generative AI (face-swap and voice-clone software) to produce illicit, non-consensual explicit content. The AI’s development and use have directly led to violations of personal rights and reputational harm, qualifying it as an AI Incident.

An AI system (face-swap/deepfake software) was directly used to create illicit content without consent, causing harm to the individual’s rights and reputation. This constitutes a realized harm under AI Incident criteria (violation of fundamental and personal rights).

No specific misuse or harm has been reported yet, but the service clearly enables creation of undetectable deepfakes that could be used for fraud, impersonation, defamation or other malicious ends. This represents a plausible risk of significant harm if deployed, making it an AI Hazard.

The deepfake services described (custom AI face‐swap and voice synthesis) were actively used to perpetrate scams—most notably the 200 million HKD fraud via a fake video conference and subsequent cases where employees transferred hundreds of thousands of yuan. These are direct harms caused by the misuse of AI systems, fitting the definition of AI Incidents.

The piece describes multiple instances where AI‐generated fake video calls and livestreams have directly led to fraud (elderly victims defrauded), infringement of celebrities’ likenesses, and harm to consumers. This constitutes realized harm caused by AI misuse, fitting the definition of an AI Incident.

The piece documents actual harms—large-scale financial fraud—directly enabled by AI face-swapping (DeepFake) technology. These events represent realized AI-driven incidents causing property and economic harm.

Although the article recounts actual incidents of deepfake scams and impersonations, its primary purpose is to describe and analyze governance proposals—legislative and technical responses—to the broader problem of AI-generated face/voice forgeries. This constitutes a societal and policy response rather than reporting a new AI incident or describing a potential hazard. Therefore, it falls under Complementary Information.

The event describes real‐world misuse of AI (deepfake face replacement) to commit fraud in e‐commerce. Consumers suffer property/financial losses due to the malicious AI application. This is a realized harm directly linked to an AI system’s misuse, so it qualifies as an AI Incident.

Although the article describes real harms from AI deepfakes (celebrity impersonation scams), its primary focus is on governance responses—legislative calls, new‐thinking approaches, and regulatory innovations—rather than reporting a new incident or hazard. Thus it provides contextual, policy‐oriented Complementary Information about ongoing efforts to address those harms.

This is a clear case of an AI system’s use (deepfake face-swap and voice-clone technology) directly causing harm (fraud and financial loss). The harms have already materialized, making this an AI Incident.

This piece provides legal interpretation and warning about AI deepfake misuse rather than reporting a specific newly occurring incident or outlining a future risk scenario. It serves to update readers on laws and enforcement perspectives, making it Complementary Information.

The misuse of AI deepfake tools is directly causing harm by enabling scams, infringing celebrity image and intellectual property rights, and deceiving consumers. These are realized harms stemming from the development and use of AI systems for malicious purposes, fitting the definition of an AI Incident.

This is a case of actual, unauthorized use of AI systems (deepfake audio/video generators) directly resulting in harm: violations of individuals’ personality and voice rights (civil and intellectual property rights), and facilitating scams. Since the AI’s misuse has led to realized legal and reputational harms, it constitutes an AI Incident.

Rather than reporting a single new incident or forecasting a specific future threat, the piece summarizes multiple infringement cases, ongoing governmental regulations, recent court rulings and expert recommendations for detecting and preventing AI deep-fake abuses. Its primary focus is on legal and policy responses and broader context, making it Complementary Information.

AI voice‐cloning systems are explicitly described and have already been used maliciously in multiple scam cases (e.g., impersonating a grandson to defraud an elderly person of tens of thousands of yuan). This constitutes an AI system’s misuse causing direct harm to individuals, meeting the criteria for an AI Incident.

The article describes concrete instances where AI deepfake tools have been used to impersonate public figures (Lei Jun, Andy Lau) and ordinary people for fraudulent schemes and unauthorized promotion. These actions constitute infringements of personal rights and have been used to scam consumers, representing direct harm caused by the use of AI systems. Thus, this is an AI Incident.

This describes actual harms—unauthorized AI-generated celebrity voices and face-swap videos used for fraud, infringement of likeness and voice rights, and resulting legal violations. The AI systems’ use directly led to these violations, fitting the definition of an AI Incident (violations of human and intellectual property rights).

While the text cites real instances of voice-cloning abuse and a recent court decision, its primary focus is on summarizing the phenomenon and advocating for policy, regulatory, and technical responses rather than reporting a single, discrete harm event. This aligns with ‘Complementary Information,’ since it offers contextual background, governance discussion, and legal developments regarding AI voice cloning.

This describes multiple incidents where AI systems for deepfake audio/video were developed and used without authorization, directly causing rights violations and fraud losses. These are concrete harms from misuse of AI, so it is an AI Incident.

This is an AI Incident because it describes actual harms—unauthorized AI‐generated deepfakes violating individuals’ legal rights and enabling fraud schemes that defrauded victims of billions. An AI system’s misuse directly led to these infringements and economic harms.

Deep‐synthesis AI systems are actively being used to produce and distribute infringing audio/video of public figures without consent, leading to rights violations (intellectual property and publicity rights) and consumer deception. This constitutes direct harm from the use of AI systems.

The described AI system (deep‐fake audio/video generators) has been used to create unauthorized celebrity content, directly violating legal rights and causing reputational and economic harm. These harms have materialized—consumers are deceived, and celebrities’ rights are infringed—meeting the criteria for an AI Incident (specifically a violation of human and intellectual property rights).

The report details actual instances where AI deep synthesis software is used to produce and distribute celebrity voice and face clones without consent, infringing on personal and intellectual property rights. This misuse of AI has directly led to legal cases and recognized harm, fitting the definition of an AI Incident (violation of human rights/intellectual property).

While the article references real harms caused by AI-generated voice impersonation (scams, misinformation, consumer deception), its primary focus is on advocating legal regulation and platform governance measures to address and prevent these abuses. This aligns with a societal/governance response to AI issues rather than reporting a single new incident or warning of a future hazard.

The piece details multiple realized harms directly resulting from AI systems (deepface/deepvoice tools): infringement of individuals’ personality and publicity rights; a broken Hong Kong fraud ring that used AI deepfakes to steal over HK$3.6 billion; a Chinese court ruling on an AI voice infringement case; and UNODC data showing a surge in AI‐enabled scam tools. These are actual incidents of harm caused by AI misuse, not speculative risks or mere background context.

The report documents realized harms from AI systems: unauthorized deep synthesis of celebrity audio-video used for scams and profit, constituting violations of rights under civil and regulatory frameworks. These are direct incidents where the development and use of AI deepfake tools have already led to infringement and fraud. Thus, it qualifies as an AI Incident.

While the article references real harms from deepfake scams and privacy violations, its primary focus is on raising awareness, highlighting existing regulations and detection techniques, and advising on multi-level prevention strategies. It serves to contextualize and update stakeholders on policy and governance responses rather than documenting a single new incident or emerging hazard.

The article describes actual harms from AI deep synthesis technology: criminal use to impersonate individuals (e.g., entrepreneurs, doctors), perpetrate fraud, and violate portrait/voice rights. These constitute realized harms directly resulting from AI misuse, fitting the AI Incident definition.

This is an actual, ongoing series of harms caused by AI systems (deep synthesis of celebrity voices and likenesses) that infringe on rights and facilitate scams. The unauthorized AI-generated content directly leads to legal violations and consumer harm, fitting the definition of an AI Incident.

The piece describes multiple realized harms directly caused by the use of AI systems for deep synthesis of voices and faces—celebrity impersonations, fraud schemes, and legal infringement cases—meeting the definition of AI Incident (violations of rights, intellectual property infringements, and fraud).

The piece does not detail a singular incident or a specific near-miss but rather surveys recurring harms from AI-synthesized voices and focuses on policy responses and regulatory needs. As it primarily provides governance context and recommendations, it qualifies as Complementary Information.

The piece focuses on regulatory and technical responses to AI voice synthesis misuse and broader governance proposals, rather than detailing a particular incident or new threat. It therefore constitutes Complementary Information, offering contextual updates on policy and oversight developments in the AI ecosystem.

This is a case of actual harm: AI voice-cloning systems are being used without consent to impersonate public figures, spreading false and offensive content that infringes on personal rights and misleads audiences. The misuse of these AI systems has directly caused legal, social and informational harm, making it an AI Incident.

No concrete harm or specific misuse event is described; instead, the article centers on the potential risks (deepfakes aiding crime, misleading information) and industry/government responses to shore up defenses. This forward‐looking discussion of plausible threats classifies it as an AI Hazard.

文中描述AI深度合成系统被用于未经授权制作名人音视频并公开售卖，直接导致侵犯名人及他人信息权利，属于AI系统的恶意或违规使用并已造成侵权伤害，符合“AI Incident”定义。

AI systems are being used to fabricate and manipulate video content, impersonating real individuals to promote worthless or potentially harmful products. Consumers have been misled into purchasing ineffective health supplements and delaying proper treatment. This misuse directly results in economic and health harms and infringes on public figures’ personality rights, fitting the definition of an AI Incident.

The article explicitly mentions AI voice synthesis systems capable of cloning voices rapidly, which qualifies as AI systems. It highlights the misuse of these AI-generated voices, implying potential harms such as fraud, misinformation, or privacy violations, although no specific incident of harm is described. Since the article focuses on the potential for misuse and the need to prevent harm rather than reporting a realized harm event, it describes a plausible risk scenario where AI voice cloning could lead to harm. Therefore, this event is best classified as an AI Hazard, as it concerns circumstances where AI use could plausibly lead to harm but no specific incident is reported.