Investigation Reports Unauthorized Deepfake Pornography Harms Thousands of Celebrities

A Channel 4 News investigation alleges that nearly 4,000 celebrities globally, including 255 British figures, were victims of deepfake pornography. Faces were superimposed onto explicit content using AI, with the top deepfake sites garnering 100 million views in three months, according to their findings.

Risk classification

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

The incident involves the creation of humiliating and sexual synthetic imagery targeting specific individuals without their consent for harassment and degradation.

Additional risk subdomains

• 1.2 Exposure to toxic content: The deepfake pornographic videos were hosted on public websites, receiving over 100 million views and exposing the public to toxic, non-consensual explicit content.

Causal factors

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

The harm is caused by human users intentionally using deployed AI deepfake generation tools to create and share non-consensual explicit imagery.

EU AI Act risk tier

• Risk tier: 3 Limited Risk

Risk Level 3. Limited Risk: The report describes AI-generated content such as deepfakes, which pose a moderate risk and require specific transparency obligations under the Act.

AI system and alleged parties

• AI system: unspecified
• AI purpose: Deepfake Video Generation; Deepfake Image Generation
• Behaviour type: Tool
• Alleged developer: Unknown deepfake technology developers
• Alleged deployer: Deepfake website operators
• Alleged harmed parties: celebrities, Cathy Newman, British public figures

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

Malicious content

Reported: Yes, the report explicitly describes the creation and spread of non-consensual deepfake pornography.

Directly caused: AI was used to superimpose faces of famous individuals onto pornographic material, creating thousands of deepfake pornographic videos.

Indirectly caused: These videos were hosted on websites that received 100 million views in three months, widely disseminating the toxic content.

Inferred additional harm: It is highly likely that millions of additional deepfake pornographic images and videos continue to be generated and spread across various online platforms, exposing millions of viewers to non-consensual explicit content.

Differential treatment

Reported: Yes, the report notes a gendered dimension to the harm.

Directly caused: The deepfake pornography disproportionately targets women, with a victim stating it is like women do not mean anything, we are just worthless, we are just a piece of meat.

Inferred additional harm: The systemic targeting of women for non-consensual sexualization reinforces gender-based harassment and unequal safety outcomes online.

Civil rights

Reported: Yes, the report describes the violation of victims' personal dignity and bodily autonomy.

Directly caused: The creation of non-consensual sexual imagery violates the fundamental right to personal dignity, privacy, and self-determination of the affected individuals.

Inferred additional harm: The widespread availability of these tools threatens the civil rights of women and public figures to participate in public life without facing targeted sexual harassment.

Psychological

Reported: Yes, the report explicitly describes psychological distress and feelings of violation and degradation.

Directly caused: Victims described feeling violated, degraded, and experiencing a loss of trust, with one stating it feels really sinister and another describing it as very violent and degrading.

Inferred additional harm: Given the 4,000 reported victims and potentially hundreds of thousands of others, widespread severe psychological distress, anxiety, and trauma are highly likely to have been caused to these individuals.

Epistemic

Reported: Yes, the report describes the fabrication of fake explicit images/videos.

Directly caused: The AI systems generated fabricated nude images and videos that falsely depicted individuals participating in pornographic acts.

Inferred additional harm: The proliferation of highly realistic deepfakes erodes public trust in visual evidence and the authenticity of digital media.

People affected

• Occurrences reported: 1
• People reportedly harmed: 4000
• People reportedly exposed: 4000

Potential causes

Management

• Slow Platform Response: Tech firms failed to proactively block deepfake ads and accounts.
• Inadequate Safety Safeguards: Platforms failed to build scalable tools to protect users from search indexing.

Technology

• AI Face Superimposition Tech: AI tools allow easy superimposition of faces onto adult content.
• Widespread App Availability: Deepfake creation apps remain widely available on various app stores.
• Rapid Technical Proliferation: Exponential growth in deepfake uploads makes detection challenging.

Data Inputs

• Unrestricted Image Harvesting: Public photos of celebrities and individuals are harvested without consent.
• Lack of Input Verification: AI models do not verify if consent was given for the input images used.

Human Factors

• Malicious Intent of Creators: Individuals create and share non-consensual porn to degrade victims.
• High Consumer Demand: Deepfake websites received over 100 million views in three months.

Process and Methods

• Reactive Removal Processes: Platforms rely on victims requesting content removal rather than preventing it.
• Search Engine Indexing: Search engines index and display deepfake sites in search results.
• Inadequate Ad Screening: Social platforms allowed ads promoting deepfake creation apps.

Regulatory Environment

• Legal Loopholes in Creation: UK law bans sharing deepfakes but does not criminalize their creation.
• Delayed Regulatory Enforcement: Ofcom rules under the Online Safety Act are not yet fully in force.

Information quality

• Classification confidence: High
• Reason for confidence: The report provides clear, detailed statistics from a Channel 4 investigation regarding the scale of deepfake pornography, specific numbers of victims, and direct quotes from affected individuals, allowing for highly confident classifications.

An investigation revealed widespread misuse of AI tools to generate non-consensual deepfake pornography targeting nearly 4,000 public figures globally. While representing a severe violation of personal privacy and showcasing advanced synthetic media capabilities, the incident does not directly threaten core national security functions, critical infrastructure, or state sovereignty.

• Overall national security impact: Minor
• Scope: Multiple nations
• Primary target: No clear primary
• Other affected: Global
• Alleged perpetrator: Unknown

Threat characteristics

• Imminence: Long-term. Represents an ongoing, strategic challenge regarding the regulation and misuse of synthetic media tools rather than an immediate national security crisis.
• Autonomy: Human-controlled. The AI systems function as tools directly operated by human users who intentionally generate and upload the explicit content.
• Novelty: Evolved capability. Represents a significant advancement in the scale, accessibility, and realism of existing deepfake technology, though the threat type itself is established.

Impact by dimension

• Physical security: Negligible. No physical systems, critical infrastructure, or kinetic targeting systems were affected by this incident.
• Information security: Minor. Demonstrates large-scale capability to generate and distribute highly realistic deepfakes of public figures, though currently utilized for non-consensual pornography rather than state-sponsored intelligence or disinformation operations.
• Sovereignty: Negligible. Core government operations, electoral systems, and state authority remained unaffected by this incident.
• Economic security: Negligible. No strategic technology theft, financial system attacks, or disruption of critical supply chains occurred.
• Societal stability: Minor. Represents a widespread violation of personal dignity and privacy targeting thousands of public figures, but impacts are manageable through standard legal and regulatory frameworks like the UK Online Safety Act.