A Law of One's Own: The Inefficacy of the DMCA for Non-Consensual Intimate Media

Content Moderation & Platform GovernanceMisinformation & Fact-CheckingPrivacy Policy Makers

Research Background and Issues

Issues and Challenges

  • Severe Harm of Non-Consensual Intimate Media (NCIM): NCIM involves the unauthorized recording, acquisition, or distribution of sexually explicit content of individuals. Victims are predominantly women (90%), often suffering severe psychological and social consequences such as anxiety, depression, and even suicidal tendencies.
  • Limitations of Existing Legal Frameworks: While nearly all U.S. states and federal laws have legislation addressing acts like "revenge porn," these laws primarily focus on penalizing the initial perpetrator rather than addressing the removal of NCIM content online. Moreover, Section 230 of the U.S. Communications Decency Act provides broad immunity to online platforms, exempting them from liability for third-party content.
  • Deficiencies in Existing Tools: The Digital Millennium Copyright Act (DMCA) is one of the few tools victims can use to request the removal of NCIM content. However, it was originally designed to protect copyright, not privacy. The process of filing a DMCA takedown request may expose the victim's personal information, and the content is often quickly re-uploaded.

Significance

  • The gendered harm caused by NCIM is not merely a personal privacy issue but also poses a threat to democratic societies, such as deepfake attacks targeting female journalists or politicians.
  • In the absence of rapid and effective platform response mechanisms, the swift dissemination of content exacerbates victims' psychological trauma and social isolation.

Research Motivation and Related Work

  • Existing research lacks empirical studies on the timeliness of NCIM reporting and removal, particularly evaluations of the effectiveness of legal tools like the DMCA in content governance.
  • The authors are motivated to fill this research gap by assessing the speed and effectiveness of the DMCA in removing NCIM content and exploring potential improvements to the current legal framework.

Proposed Solution

Proposed Approach

  • This study is the first to quantitatively evaluate the effectiveness of the DMCA in removing NCIM content, utilizing a database containing over 54,000 DMCA takedown requests (involving more than 85 million URLs) spanning over a decade.
  • The study employs "survival analysis" to measure the persistence of NCIM content on web hosts and Google search results.

Innovations

  • Metric Design: Introduced the concept of "Time-To-Takedown" (TTT) to evaluate the time required to remove content from platforms or search engines, using "48 hours" as a key benchmark for removal efficiency.
  • Comparative Analysis: Differentiated between "commercial" and "non-commercial" NCIM content, providing a basis for more targeted policy recommendations.
  • Focus on Small Platforms: The study not only examines large platforms but also analyzes the infringement characteristics of small, underregulated platforms.

Implementation Steps

  1. Data Collection: Extracted DMCA takedown records from Harvard University's Lumen database, filtering NCIM-related reports using keywords such as "revenge porn."
  2. Dual Monitoring:
    • Used HTTP status codes (e.g., "200" or "404") to track whether website content was removed.
    • Employed the Google Custom Search API to detect whether URLs were delisted from Google's index.
  3. Classification: Used artificial intelligence (GPT-4) to classify takedown requests, distinguishing between commercial and non-commercial NCIM.
  4. Statistical Analysis: Applied Kaplan-Meier survival analysis to evaluate content removal efficiency, including comparisons between commercial and non-commercial cases.

Research Findings

Key Findings

  1. Average Time-To-Takedown:

    • Web Hosts: Only 5.39% of non-commercial NCIM URLs were removed within 48 hours, with a median removal time of 45.67 days.
    • Google Search: Approximately 52.27% of non-commercial URLs were delisted within 48 hours, with a median removal time of 1.68 days.
  2. Significant Impact of Small Websites:

    • Non-commercial NCIM is primarily concentrated on low-traffic small websites, yet these sites still attract millions of monthly views, highlighting their role in facilitating the spread of NCIM content.
  3. 10-Year Longitudinal Trends:

    • With the growth of internet usage, particularly the creator economy (e.g., OnlyFans), the number of DMCA takedown requests has significantly increased.
    • Reports of commercial NCIM have steadily risen, while reports of non-commercial NCIM have remained relatively stable at about three cases per day.

Advantages Over Existing Solutions

  • This study is the first to reveal the effectiveness and limitations of the DMCA in addressing NCIM.
  • It provides a comparative analysis of content governance on large platforms versus small websites, addressing a gap in existing research that primarily focuses on major tech companies.

Limitations and Future Directions

  • Geographical Limitations: The data is primarily based on English keywords and focuses on the U.S. legal and cultural context, necessitating further research on NCIM handling in a global context.
  • Deepfake Content Limitations: AI-generated deepfake content has not been comprehensively included in this study, but future efforts should design specific regulations to address such misuse.
  • Data Scope: The Lumen database does not capture the entirety of NCIM takedown requests, as its users tend to be individuals with higher technical or legal awareness, potentially underestimating the issue's prevalence.

Conclusion and Policy Recommendations

  • Legal Revisions: Advocate for the creation of dedicated NCIM legislation rather than relying on the DMCA, which primarily serves economic interests.
  • Accountability for Small Platforms: Strengthen regulation of small and medium-sized websites, mandating that they remove content within 48 hours of receiving a complaint.
  • Accelerated Takedown Processes: Explore automated reporting and processing systems to prioritize "rapid online content hiding," reducing exposure while allowing time for review.
  • AI Legal Applications: Clarify the copyright and liability of AI-generated deepfake content as soon as possible, and establish specific regulations for new forms of NCIM.

This study provides robust data highlighting the shortcomings of existing tools and offers critical insights for future policy, technological, and legal reforms.

Quick Actions

Share

Share this page

ios_share

https://hci.top/en/papers/chi/188964/2025

AdRecommended

Learn AI Coding at CodeNow

open_in_newOpen DOI Link
DOI: https://dl.acm.org/doi/10.1145/3706598.3713334
At a Glance

Paper Snapshot

fact_check
dataset
Source
CHI
calendar_month
Year
2025
emoji_events
Award
No award tagged
group
Authors
6 authors
sell
Subtopics
Content Moderation & Platform Governance, Misinformation & Fact-Checking
work
Professions
Privacy Policy Makers
article
Content Status
Full text indexed
hub
Related Papers
4 related papers