## Heatmap: Security & Privacy Properties in AI Literature ### Overview The image is a heatmap visualizing the prevalence of different security and privacy properties discussed in AI literature across a range of publications. The heatmap uses a color gradient from light to dark blue to represent the frequency or emphasis given to each property in each publication. The rows represent security and privacy properties, while the columns represent specific publications. ### Components/Axes * **Y-axis (Rows):** Security and privacy properties, including: * Quality * Transparency * Privacy * Accuracy * Accountability * Safety * Integrity * Reliability * Security * Robustness * Availability * Usability * Authenticity * Controllability * Resilience * **X-axis (Columns):** Publications, listed with author names and publication year: * Shailya et al. 2025 * Magana & Shilton 2025 * Paraschou et al. 2025 * Kinahan et al. 2024 * Pareek et al. 2024 * Kim et al. 2024 * Toney et al. 2024 * Casper et al. 2024 * Maeda & Quanhaase 2024 * Manzini et al. 2024 * Inie et al. 2024 * Wang et al. 2024 * Alpherts et al. 2024 * Scharowski et al. 2023 * Knowles et al. 2023 * Shelby et al. 2023 * Williams & Haring 2023 * Brand 2023 * Lawrence et al. 2023 * Panigutti et al. 2023 * Schmitz 2023 * Kim et al. 2023 * Nannini et al. 2023 * Ferrario & Loi 2022 * Zilka et al. 2022 * Engelmann et al. 2022 * Thornton et al. 2022 * Schoeffer et al. 2022 * Liao & Sundar 2022 * Knowles & Richards 2021 * Kim et al. 2021 * Andrus et al. 2021 * Loi & Spielkamp 2021 * Jacovi et al. 2021 * Thornton et al. 2021 * Lakkaraju & Bastani 2020 * Huang et al. 2020 * Zhang & Dafoe 2020 * Bhatt et al. 2020 * Toreini et al. 2020 * Engelmann et al. 2019 * Mohseni 2019 * **Color Scale (Legend):** Located on the right side of the heatmap. It represents the degree to which a property is emphasized in a publication, ranging from 0.0 (lightest blue) to 1.0 (darkest blue). The scale has tick marks at 0.0, 0.2, 0.4, 0.6, 0.8, and 1.0. ### Detailed Analysis The heatmap shows varying degrees of emphasis on different security and privacy properties across the publications. Here's a breakdown of some observations: * **Quality:** Shailya et al. 2025 shows a medium emphasis (blue, ~0.5). * **Transparency:** Magana & Shilton 2025 shows a high emphasis (dark blue, ~0.8-1.0). * **Privacy:** Paraschou et al. 2025 shows a high emphasis (dark blue, ~0.8-1.0). * **Accuracy:** Kinahan et al. 2024 shows a high emphasis (dark blue, ~0.8-1.0). * **Accountability:** Pareek et al. 2024 shows a high emphasis (dark blue, ~0.8-1.0). * **Safety:** Kim et al. 2024 shows a medium emphasis (blue, ~0.5). * **Integrity:** Toney et al. 2024 shows a low emphasis (light blue, ~0.2). * **Reliability:** Casper et al. 2024 shows a medium emphasis (blue, ~0.5). * **Security:** Maeda & Quanhaase 2024 shows a high emphasis (dark blue, ~0.8-1.0). * **Robustness:** Manzini et al. 2024 shows a low emphasis (light blue, ~0.2). * **Availability:** Inie et al. 2024 shows a high emphasis (dark blue, ~0.8-1.0). * **Usability:** Wang et al. 2024 shows a low emphasis (light blue, ~0.2). * **Authenticity:** Alpherts et al. 2024 shows a low emphasis (light blue, ~0.2). * **Controllability:** Scharowski et al. 2023 shows a high emphasis (dark blue, ~0.8-1.0). * **Resilience:** Knowles et al. 2023 shows a high emphasis (dark blue, ~0.8-1.0). ### Key Observations * Some properties like "Privacy," "Accuracy," "Accountability," and "Security" appear to be frequently emphasized in the literature, as indicated by the darker blue shades in their respective rows. * Other properties like "Integrity," "Robustness," "Usability," and "Authenticity" seem to receive less attention overall, as indicated by the lighter blue shades. * There is considerable variation in the emphasis given to different properties across different publications. Some publications focus heavily on certain properties while largely ignoring others. * The publications from 2024 and 2025 appear to have a higher concentration of dark blue cells, suggesting a potentially increasing focus on these properties in more recent research. ### Interpretation The heatmap provides a visual representation of the relative importance and focus given to different security and privacy properties in AI research. The variations in color intensity suggest that certain properties are considered more critical or are more frequently addressed in the literature than others. This could reflect the evolving priorities and challenges in the field of AI security and privacy. The increasing emphasis on certain properties in more recent publications may indicate emerging trends or a growing awareness of specific risks and vulnerabilities. The heatmap can be used to identify gaps in the literature and to guide future research efforts towards addressing less-explored but potentially important security and privacy considerations.