Facial Recognition Service Abused to Target Russian Porn Actresses

The facial recognition service FindFace was utilized by users of the imageboard 'Dvach' to deanonymize and harass women working in the sex industry. Participants in the campaign matched photos of these women to their Vkontakte social media profiles and subsequently spammed the victims' friends and families with the information. The service's founder stated that they could not prevent such misuse but offered to cooperate with authorities to identify the harassers.

The facial recognition software FindFace allowing its users to match photos to people’s social media pages on Vkontakte was reportedly abused to de-anonymize and harass Russian women who appeared in pornography and alleged sex workers.

Source: AI Incident Database

Risk classification

  • Primary risk domain: 2 Privacy & Security
  • Primary risk subdomain: 2.1 Compromise of privacy by obtaining, leaking or correctly inferring sensitive information

The FindFace AI system was used to correctly infer the real-world identities of anonymous individuals (pornographic actresses and subway passengers) without their consent, compromising their privacy.

Additional risk subdomains

  • 4.3 Fraud, scams, and targeted manipulation: The AI tool was leveraged by malicious actors for the targeted humiliation, doxing, and harassment of specific victims.

Causal factors

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

The primary cause of the harm was the intentional decision and action of Dvach users to weaponize the deployed FindFace tool to dox and harass women.

EU AI Act risk tier

  • Risk tier: 2 High Risk

High Risk: The system is a facial recognition application used for biometric identification of individuals, which has significant implications for fundamental rights and privacy, similar to high-risk AI categories like biometric identification.

AI system and alleged parties

  • AI system: FindFace
  • AI purpose: Face Recognition; Identification
  • Behaviour type: Tool
  • Alleged developer: NtechLab
  • Alleged deployer: NtechLab
  • Alleged harmed parties: Russian sex workers, Russian pornographic actresses

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

Malicious content

Reported: The report explicitly describes the creation and spread of doxing files and harassing messages sent to the victims' families and friends.

Indirectly caused: Dvach users distributed archived copies of the victims' social media pages and sent spam messages to their social networks.

Inferred additional harm: It is likely that toxic and harassing messages reached hundreds of friends and family members of the targeted women.

Differential treatment

Reported: The report explicitly describes a campaign targeting Russian women who appear in pornography or are registered on prostitution services.

Indirectly caused: The harassment campaign singled out and targeted women in the sex industry due to moral outrage, subjecting them to targeted doxing and harassment.

Inferred additional harm: This targeted harassment reinforces societal stigma and unequal treatment of women in the sex industry.

Civil rights

Reported: The report explicitly describes severe privacy violations and harassment targeting specific individuals.

Indirectly caused: The right to privacy of the targeted women was severely violated when their anonymous identities were linked to their real-world profiles without consent.

Inferred additional harm: The incident represents a broader threat to the civil right to privacy in public spaces and online anonymity.

Privacy

Reported: The report explicitly describes privacy violations, noting that FindFace was used to match random photographs to people's social media pages.

Directly caused: The FindFace AI system matched photographs of individuals to their Vkontakte profiles, directly revealing their real-world identities without their consent.

Indirectly caused: Dvach users used these matches to obtain and distribute archived copies of the victims' social media pages to their families and friends.

Inferred additional harm: It is likely that thousands of other Vkontakte users had their privacy compromised or felt compelled to alter their privacy settings due to the threat of deanonymization.

Psychological

Reported: The report explicitly describes a nasty effort to identify and harass Russian women, including spamming their families and friends.

Indirectly caused: The doxing and targeted harassment campaign caused significant distress, anxiety, and social humiliation to the targeted women and their families.

Inferred additional harm: It is highly likely that the targeted women suffered severe psychological distress, anxiety, and trauma due to their private lives and identities being exposed to their social circles.

People affected

  • Occurrences reported: 1
  • People reportedly harmed: 50
  • People reportedly exposed: 100

Potential causes

Management

  • Passive Abuse Mitigation Approach: Management accepted harassment as unpreventable, relying on post-hoc policing.
  • Inadequate Risk Assessment: Failure to foresee and mitigate obvious harassment risks before release.

Technology

  • Facial Recognition Matching: Highly accurate matching of random photos to social media profiles.
  • Lack of Abuse Prevention Tech: No built-in technical blocks to prevent automated or mass reverse-searches.

Data Inputs

  • Public Social Media Photos: Unprotected Vkontakte profile pictures used as the matching database.
  • Target Images from Pornography: Images scraped from adult sites used as query inputs for matching.

Human Factors

  • Malicious Intent of Online Users: Dvach users motivated by moral outrage and resentment targeted women.
  • Inspirational Misuse Demonstrations: Public demonstrations of the tool's invasiveness inspired harassers.

Process and Methods

  • Absence of Consent Verification: No process to verify if the subject consented to being searched.
  • Inadequate User Activity Monitoring: No mechanism to flag or block users performing malicious searches.

Regulatory Environment

  • Weak Privacy Regulations: Lack of legal frameworks preventing biometric scraping of social media.

Information quality

  • Classification confidence: High
  • Reason for confidence: The report provides clear, consistent details about how the FindFace application was used, the specific doxing campaign on Dvach, and the reactions of the platform (Vkontakte) and the developer. There are no major conflicting accounts, and the role of the AI system is explicitly described.
  • Ambiguities identified: The exact number of women targeted in the doxing campaign is not specified.
  • Alternative interpretations: None.

The incident involved the weaponization of the FindFace facial recognition tool by online actors to deanonymize and harass Russian citizens. While causing severe privacy violations and psychological distress to targeted individuals, the incident's national security implications remain minor, representing a localized law enforcement and regulatory challenge rather than a systemic threat to state stability.

  • Overall national security impact: Minor
  • Scope: Single nation
  • Primary target: Russia
  • Alleged perpetrator: Users of the imageboard Dvach

Threat characteristics

  • Imminence: Long-term. The immediate crisis has passed, but the underlying capability for public deanonymization represents an ongoing strategic privacy concern.
  • Autonomy: Human-controlled. The AI system acted as a tool directly queried and manipulated by human actors to perform specific matching tasks.
  • Novelty: Evolved capability. Represented a significant advancement in the public availability and weaponization of facial recognition technology for mass deanonymization.

Impact by dimension

  • Physical security: Negligible. No physical systems, kinetic threats, or critical infrastructure were affected or targeted in this incident.
  • Information security: Minor. While the tool enabled targeted doxing and harassment, it did not involve state-sponsored information warfare, intelligence compromise, or classified data theft.
  • Sovereignty: Negligible. The incident did not compromise government decision-making, electoral systems, or core state authority.
  • Economic security: Negligible. There was no threat to financial systems, strategic technological assets, or critical economic infrastructure.
  • Societal stability: Minor. The incident represents a significant privacy violation and targeted harassment campaign against citizens, but its societal impact was localized and manageable through standard law enforcement.
Explore in the interactive Incident Tracker