Mental Models, Expectations and Implications of Client-Side Scanning: An Interview Study with Experts

Literature Title

Mental Models, Expectations, and Implications of Client-Side Scanning: An Interview Study with Experts

Literature Information

• Subject Area: Information security and privacy protection, specifically the application of client-side scanning (CSS) technology in preventing the dissemination of child sexual abuse material (CSAM).
• Keywords: Client-Side Scanning (CSS), Child Sexual Abuse Material (CSAM), Privacy Protection, Security and Privacy, Crime Prevention.

Research Background and Issues

• Identified Problems or Challenges:

  1. Under the current legal and technical frameworks, the widespread use of encrypted communication has made traditional server-side scanning difficult to implement, while client-side scanning (CSS) is viewed as a potential solution.
  2. CSS is a deeply socially contextualized issue, with significant disagreements among stakeholders regarding the technology, its application scenarios, and potential societal impacts.
  3. Current discussions on CSS lack a comprehensive definition, and there are inconsistencies in understanding its technical capabilities, expected applications, and potential impacts.

• Significance:

  • This issue involves balancing privacy and rights with preventing the dissemination of CSAM, making it a globally significant and highly sensitive topic.
  • A clear definition of CSS and an understanding of its potential societal interactions are crucial for shaping future social and technological policies.

• Research Motivation and Preliminary Basis:

  • To explore the definition of CSS and the mental models of experts on this issue, filling the research gap and providing a foundation for interdisciplinary discussions.
  • The data source consists of semi-structured interviews with 28 experts from various fields, including child protection, law, privacy protection, and technology companies.

Solution

• Methods and Research Objectives:

  • The authors conducted qualitative interview research, analyzing the conceptual understanding, expectations, and potential impacts of CSS as perceived by 28 domain experts, and summarized their mental models.
  • The main research objectives include:
    1. Establishing and analyzing experts' multidimensional understanding of CSS (how it works, what content is scanned, etc.).
    2. Exploring their expectations for CSS technology and its practical challenges.
    3. Analyzing the potential impacts of CSS on individual and societal levels.

• Research Innovations:

  1. Structuring CSS into a data-driven framework, clarifying its key technical dimensions (input, output, algorithms, user interaction, reporting, etc.).
  2. Proposing universally applicable mental model dimensions for CSS to facilitate discussions on technology, privacy, and ethics among stakeholders.
  3. Incorporating diverse stakeholder perspectives, including privacy preservation interests and the feasibility of technical implementation, to achieve a balanced discussion.

• Implementation Steps and Key Techniques:

  • Designed a detailed, multidimensional interview framework, collecting expert data related to structured goals through open-ended questions.
  • Visualized experts' abstract understanding of CSS using flowcharts and event sequence analysis, forming technical mental models.
  • Employed a hybrid coding method for qualitative analysis (including open coding and selective coding).

Research Findings

• Specific Findings:

  • Extracted a mental model framework for CSS, including:
    1. Inputs during system construction (e.g., scanning algorithms, triggering conditions, operational levels).
    2. Key outputs (e.g., match results, false positive possibilities).
    3. User interaction aspects (e.g., whether users are notified, whether user consent is required).
    4. Interaction with third-party reporting modules (e.g., minimum reporting thresholds).
  • Identified technical and societal risks, such as potential false positives, trust risks due to lack of transparency, and the danger of misuse for expanded scanning purposes.

• Comparative Advantages Over Existing Solutions:

  1. In global privacy debates, CSS is the only method that enables pre-scanning on the client side while preserving the integrity of E2EE (end-to-end encryption).
  2. The study goes beyond a single technical discussion, using multiple case scenarios (CSAM, copyright protection, advertising) to comprehensively evaluate the risks of technical implementation.

• Experimental or Validation Results:

  • Expert input validated the cross-domain complexity of CSS, highlighting the difficulty of establishing a single definition and pointing out the need for different implementation norms in varying social contexts.

• Limitations and Future Directions:

  • Limitations:
    1. Most participants were German experts, introducing a cultural bias in the sample.
    2. Data was based solely on interviews, without experimental validation, making it impossible to quantify the technology's effectiveness.
    3. The authors' backgrounds in information technology may have introduced a bias favoring privacy protection.
  • Future Directions:
    1. Explore more advanced non-invasive technologies, such as heuristic encryption-based non-intrusive scanning.
    2. Expand the scope of policy-level discussions to more diverse regions and include the perspectives of non-technical experts.
    3. Design "low-intrusion" CSS for addressing societal issues beyond child protection.

Conclusion

This study, through the perspectives of experts from multiple fields, provides a foundational analytical framework for the feasibility, societal risks, and ethical trade-offs of CSS technology. Despite its potential privacy risks and societal controversies, its promise in combating major crimes such as CSAM warrants continued exploration.