Non-Consensual Synthetic Intimate Imagery: Prevalence, Attitudes, and Knowledge in 10 Countries

Title of the Paper

Non-Consensual Synthetic Intimate Imagery: Prevalence, Attitudes, and Knowledge in 10 Countries

Paper Information

• Topic Area: Social impacts, legal, and technical responses to non-consensual synthetic intimate imagery (NSII) and deepfake pornography.
• Keywords: Deepfake pornography, deepfake, AI-generated imagery, image-based sexual abuse, non-consensual explicit imagery, synthetic pornography

Research Background and Issues

• Issues or Challenges:
  • The misuse of deepfake technology, particularly for creating non-consensual synthetic intimate imagery (NSII), is exacerbating issues of online gender-based violence and sexual abuse.
  • Current legal frameworks, detection technologies, and societal understanding are insufficient to address AI-generated image-based sexual abuse (AI-IBSA).
• Significance:
  • Deepfake pornography is widely used for harassment, sexual extortion, and humiliation, with increasingly severe social, legal, and psychological consequences.
  • The low barrier to misuse of synthetic imagery means virtually anyone can become a victim.
• Research Motivation and Related Work:
  • The authors aim to address gaps in research on the prevalence, public attitudes, and gender differences related to victimization and perpetration of deepfake pornography through a cross-national study.
  • Cited related studies explore societal anxieties surrounding deepfake technology, inadequate detection methods, and legislative challenges associated with gender-based violence.

Solution

• Proposed Methods or Solutions:

  • Survey Study: The authors conducted an online survey of over 16,000 adult respondents across 10 countries to understand public awareness, attitudes, and data on victimization and perpetration related to AI-IBSA.
  • Data Analysis: Quantitative analysis and qualitative data coding were applied to uncover geographical differences, gender impacts, and trends in deepfake phenomena.

• Innovations:

  • The first systematic evaluation of the prevalence and societal attitudes toward AI-generated pornography across 10 countries.
  • Comparison of "traditional" image-based sexual abuse behaviors with AI-generated synthetic behaviors.
  • Recommendations for more effective policy interventions in the context of limited public awareness.

• Implementation Steps:

  1. Design a survey questionnaire targeting AI-IBSA issues.
  2. Analyze differences in awareness and legislative contexts across countries.
  3. Assess prevalence rates of victimization and perpetration, comparing gender and cultural differences.
  4. Provide recommendations for policy, education, and technological solutions.

Research Findings

• Specific Findings:

  1. Low Awareness of Deepfake Pornography: On average, only 28% of respondents had heard of deepfake pornography.
  2. Negative Public Attitudes Toward Non-Consensual Synthetic Content: Most respondents believed the creation or distribution of NSII should be criminalized.
  3. Relatively Rare Victimization and Perpetration: Overall, 2.2% of respondents reported experiencing AI-IBSA victimization, while 1.8% admitted to perpetration.
  4. Significant Gender Differences: Men were more likely to perceive these behaviors as less harmful and were more likely to report experiences of viewing, victimization, and perpetration.
  5. Limited Impact of Legislation: Even in countries with clear laws (e.g., Australia, South Korea), victimization and perpetration rates did not significantly decrease.

• Advantages Over Existing Solutions:

  • Provides detailed cross-national and gender-specific data to address existing research gaps.
  • Highlights deficiencies in current technologies and legal frameworks for preventing deepfake pornography-related behaviors.

• Experimental or Evaluation Results:

  • Deepfake pornography content is widely perceived as "harmful" (regardless of whether the target is a celebrity or an ordinary individual).
  • Consumption of deepfake content is the most common behavior, followed by more active creation or distribution behaviors.

• Limitations and Future Directions:

  • Limitations: Respondents may underestimate or lack awareness of their victimization by deepfake imagery; some perpetration behaviors may go unreported.
  • Suggestions for future research include:
    1. In-depth exploration of motivations behind AI-IBSA perpetration (e.g., sexual extortion, technological demonstration).
    2. Development of efficient detection technologies to reduce the spread of such content.
    3. Implementation of gender-focused educational interventions.
    4. Expansion of survey scope to include more developing countries or specific victim groups.

Conclusion

This study provides a forward-looking perspective by conducting an in-depth survey of respondents from 10 countries, revealing the urgent need for legal, technological, and educational interventions to address AI-generated deepfake content. It emphasizes the importance of gender factors, cultural contexts, and societal awareness, pointing to potential pathways for preventing AI-based gender violence in the future.