A Scoping Review and Guidelines on Privacy Policy's Visualization from an HCI Perspective

Shuning Zhang, Eve He, Sixing Tao, Yuting Yang, Ying Ma, Ailei Wang, Xin Yi, Hewu Li

TL;DR

This paper provides a comprehensive scoping review of 65 top-tier research papers tracking the evolution of privacy policy visualizations over the past two decades. It maps how the field has shifted from static, text-heavy disclosures to interactive, AI-driven agents, highlighting the critical tensions between information load, design complexity, and stakeholder incentives.

What Problem Does This Paper Address?

Privacy policies are the cornerstone of informed consent, yet they are notoriously ineffective in practice. Usually presented as lengthy, jargon-filled legal documents, they are routinely ignored by users, leading to a state of "informed helplessness."

While Human-Computer Interaction (HCI) researchers have spent decades designing visual alternatives—such as privacy "nutrition labels," interactive dashboards, and comic-based narratives—these solutions rarely see widespread real-world adoption. This paper addresses the fragmented landscape of privacy policy visualization by analyzing its historical trajectory through a user-centered design lens to uncover why past solutions failed and how modern technologies like Large Language Models (LLMs) can chart a path forward.

Methods

The authors conducted a systematic scoping review following the PRISMA-ScR guidelines. They screened 719 papers from top-tier HCI and security venues (such as CHI, SOUPS, and USENIX Security) published since 2000, ultimately selecting a curated corpus of 65 papers.

To synthesize the literature, they developed a multi-dimensional analytical framework adapted from the user-centered design lifecycle. This framework organizes the domain's evolution across four interconnected dimensions: Understanding Context, Establishing Requirements, Designing Solutions, and Evaluation.

The framework of this paper, organized around context, requirements, design, and evaluation.

Datasets

As a scoping review, the primary dataset analyzed is the corpus of 65 academic papers spanning from 2000 to 2025 (fully detailed in the paper's appendix). However, the review highlights several specific datasets and benchmarks used within the analyzed literature, such as:

  • CA4P-483: A dataset constructed to address regional compliance with Chinese cybersecurity laws alongside GDPR and CCPA.
  • GenAIPABench: An evaluation benchmark used to test the reliability and performance of generative AI privacy assistants.

Key Insights and Ideas

  • The Co-Evolutionary Spiral of Design and Automation: Advanced design concepts (like context-aware or conversational privacy notices) initially stalled because manually populating them was unscalable. Conversely, breakthroughs in NLP and LLMs are now unlocking the exact semantic interpretation needed to make these complex designs viable.
  • From Disclosure to Delegation: The paradigm is shifting from simply summarizing text (passive reading) to proactive, AI-driven agents. Future systems are moving toward "closing the loop" by coupling privacy awareness with delegated execution—allowing AI agents to automatically negotiate privacy settings or execute data deletion requests on the user's behalf.
  • The Generality vs. Specificity Dilemma: While standardized formats (like nutrition labels) offer cross-platform familiarity, the rise of screenless IoT devices and immersive VR/AR environments demands highly specialized, multimodal, and situated interaction patterns (e.g., haptic feedback or spatial overlays).
  • Upstream Incentive Misalignment: The lack of real-world adoption isn't just a design failure; it's an economic and operational one. Developers often view privacy compliance as a secondary burden divorced from core engineering tasks, while data-monetizing business models actively disincentivize clear, legible disclosures.

Key Results

The review maps out a clear temporal shift in publication trends, showing a significant surge in privacy policy visualization research that peaked between 2020 and 2023, with CHI serving as the primary venue (25 out of 65 papers).

Qualitatively, the paper synthesizes the evolution of design solutions into a clear timeline: starting with machine-readable protocols (P3P) in the early 2000s, moving to tabular nutrition labels and NLP extraction in the 2010s, and culminating in LLM-powered interactive dialogues today. The authors distill these insights into a comprehensive mapping table linking four major open challenges (Engagement, Device Contexts, Demographics, and Production Integration) to actionable design guidelines for the HCI community.

Limitations and Future Work

The authors note that their review strictly limits its scope to the visualization of the policy content itself, excluding broader operational privacy controls (like simple toggle switches) or general privacy attitudes.

For future work, they outline four critical open challenges:

  1. Improving Engagement: Moving beyond static displays to context-aware dynamic visualizations and "subtle interventions" (like policy-learning CAPTCHAs) to stimulate reflective thinking.
  2. Tailoring to Diverse Devices: Developing ambient and multimodal delivery channels for screenless IoT and native policy integrations for VR/AR spaces.
  3. Customizing for Demographics: Leveraging LLMs to dynamically transcode legal text into simplified modalities tailored to vulnerable groups (e.g., children or visually impaired users) without intrusive profiling.
  4. Production Integration: Creating developer-centric automation tools, such as AI-powered IDE co-pilots, to lower the cognitive burden on developers and automatically scaffold privacy interfaces during software development.

Summary updated

Abstract

Privacy Policies are a cornerstone of informed consent, yet a persistent gap exists between their legal intent and practical efficacy. Despite decades of Human-Computer Interaction (HCI) research proposing various visualizations, user comprehension remains low, and designs rarely see widespread adoption. To understand this landscape and chart a path forward, we synthesized 65 top-tier papers using a framework adapted from the user-centered design lifecycle. Our analysis presented findings of the field's evolution across four dimensions: (1) the trade-off between information load and decision efficacy, which demonstrates a shift from augmenting disclosures to prioritizing information condensation and cognitive load management to counter the inefficacy of comprehensive texts, (2) the co-evolutionary dynamic of design and automation, revealing that complex design ambitions such as context-awareness drove the need for advanced NLP, while recent LLM breakthroughs are enabling the semantic interpretation required to realize those designs, (3) the tension between generality and specificity, highlighting the divergence between standardized, cross-platform solutions and the increasing necessity for specialized, context-aware interaction patterns in IoT and immersive environments, and (4) balancing stakeholder opinions, which shows that visualization efficacy is constrained by the complex interplay of regulatory mandates, developer capabilities and provider incentives. We conclude by outlining four critical challenges for future research.
BibTeX
@inproceedings{zhang2026scoping,
  author = {Shuning Zhang and Eve He and Sixing Tao and Yuting Yang and Ying Ma and Ailei Wang and Xin Yi and Hewu Li},
  title = {{A Scoping Review and Guidelines on Privacy Policy's Visualization from an HCI Perspective}},
  booktitle = {CHI},
  year = {2026},
  doi = {10.1145/3772318.3790320},
}
Figure