Are You Comfortable Sharing It?: Leveraging Image Obfuscation Techniques to Enhance Sharing Privacy for Blind and Visually Impaired Users Conference: Proceedings of the 2026 CHI Conference on Human Factors in Computing Systems; April 13–17, 2026; Barcelona, Spain Proceedings of the 2026 CHI Conference on Human Factors in Computing Systems (CHI ’26), April 13–17, 2026, Barcelona, Spain DOI: 10.1145/3772318.3791916 ISBN: 979-8-4007-2278-3/2026/04 CCS: Human-centered computing Accessibility design and evaluation methods CCS: Human-centered computing User studies CCS: Information systems Multimedia content creation
Satabdi Das, Nahian Beente Firuj, Manjot Singh, Arshad Nasser, Khalad Hasan · 2026
Read Are You Comfortable Sharing It?: Leveraging Image Obfuscation Techniques to Enhance Sharing Privacy for Blind and Visually Impaired Users Conference: Proceedings of the 2026 CHI Conference on Human Factors in Computing Systems; April 13–17, 2026; Barcelona, Spain Proceedings of the 2026 CHI Conference on Human Factors in Computing Systems (CHI ’26), April 13–17, 2026, Barcelona, Spain DOI: 10.1145/3772318.3791916 ISBN: 979-8-4007-2278-3/2026/04 CCS: Human-centered computing Accessibility design and evaluation methods CCS: Human-centered computing User studies CCS: Information systems Multimedia content creation by Satabdi Das, Nahian Beente Firuj, Manjot Singh, Arshad Nasser, Khalad Hasan (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2603-03606.txt.
Introduction
Satabdi Das and four colleagues ask blind and visually impaired people what would make photo sharing feel safe. A shared image can accidentally expose sensitive or embarrassing content its owner cannot see, risking privacy and relationships. Their 2026 study tests image obfuscation filters as a pre-sharing shield, with twenty blind and visually impaired participants judging filtered images of varying sensitivity.
Pixelation was the least preferred filter. Other preferences shifted with image type and with the audience, family, friends, or strangers, so no single filter wins everywhere. Across audiences, participants felt more comfortable sharing filtered images than unfiltered ones. The authors turn those patterns into design guidelines for sharing tools that warn and protect without taking control away from the person sharing.
Catalog record
| CREATOR | Satabdi Das, Nahian Beente Firuj, Manjot Singh, Arshad Nasser, Khalad Hasan |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 54354 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2603.03606
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Are You Comfortable Sharing It?: Leveraging Image Obfuscation Techniques to Enhance Sharing Privacy for Blind and Visually Impaired Users Conference: Proceedings of the 2026 CHI Conference on Human Factors in Computing Systems; April 13–17, 2026; Barcelona, Spain Proceedings of the 2026 CHI Conference on Human Factors in Computing Systems (CHI ’26), April 13–17, 2026, Barcelona, Spain DOI: 10.1145/3772318.3791916 ISBN: 979-8-4007-2278-3/2026/04 CCS: Human-centered computing Accessibility design and evaluation methods CCS: Human-centered computing User studies CCS: Information systems Multimedia content creation Satabdi Das, Nahian Beente Firuj, Manjot Singh, Arshad Nasser, Khalad Hasan Abstract . People with Blind Visual Impairments (BVI) face unique challenges when sharing images, as these may accidentally contain sensitive or inappropriate content. In many instances, they are unaware of the potential risks associated with sharing such content, which can compromise their privacy and interpersonal relationships. To address this issue, we investigated image filtering techniques that could help BVI users manage sensitive content before sharing with various audiences, including family, friends, or strangers. We conducted a study with 20 BVI participants, evaluating different filters applied to images varying in sensitivity, such as personal moments or embarrassing shots. Results indicated that pixelation was the least preferred method, while preferences for other filters varied depending on image type and sharing context. Additionally, participants reported greater comfort when sharing filtered versus unfiltered images across audiences. Based on the results, we offer a set of design guidelines to enhance the image-sharing experience for BVI individuals. † † cc-license: by 1. Introduction Blind and Visually Impaired (BVI) individuals often share photos with others to access visual information - such as by requesting descriptions of the image content from sighted peers or volunteers to understand objects, people, text, or scenes depicted in the photos ( 56 ; 11 ; 18 ) . They also use images to socialize, for example, by sharing moments from their lives with friends and family, to preserve memories, and express themselves creatively ( 33 ; 39 ; 50 ) . In addition, mobile cameras and computer vision enable ‘visually aware’ applications to assist BVI with everyday tasks, such as identifying objects, recognizing faces, and reading text ( 5 ) . For instance, they utilize Visual Description Services (VDS) via phone cameras, which leverage technologies such as computer vision and artificial intelligence to perform tasks like recognizing objects, reading text, and identifying facial expressions. VDS fall into AI-powered (e.g., Seeing AI ( 46 ) ), human-powered (e.g., Be My Eyes ( 10 ) ), and hybrid approaches that combine both (e.g., VizWiz ( 13 ) ). However, this convenience introduces privacy challenges, as BVI users may unintentionally capture and share images containing sensitive or inappropriate content (e.g., private documents, identifiable individuals, or revealing backgrounds) without being aware of it. As a result, their privacy may be compromised, such as by inadvertently disclosing personal or third-party information, and this can potentially strain interpersonal relationships due to perceived breaches of trust or unintended oversharing ( 39 ; 52 ) . Addressing these privacy concerns is essential to empower BVI users to confidently and safely share photos without risking their own or others’ privacy. Recent research has explored privacy-enhancing technologies to support blind individuals in managing the visual content they share ( 7 ; 48 ; 29 ) . A promising approach involves applying computer vision techniques to automatically detect and obscure sensitive image elements - such as through blurring, masking, or removal ( 8 ; 23 ; 37 ; 54 ; 16 ) . These studies highlight that blind users value having agency over how privacy is maintained in their photos and advocate for tools that are designed with accessibility in mind to support informed decision-making ( 7 ; 48 ; 25 ) . Although recent work has advanced privacy-preserving tools for BVI individuals, existing approaches reveal several gaps that motivate our work. Alharbi et al. ( 7 ) explored speculative obfuscation in visual description services but did not compare different filtering techniques across sensitive content categories. Zhang et al. ( 55 ) proposed a human-AI hybrid workflow for detecting and redacting private information, yet focused on usability rather than comparing obfuscation across diverse content types. Similarly, prior work on computer-generated descriptions ( 54 ) considered sense-making and trust but did not explore how obfuscation affects BVI users’ comfort sharing images with different audiences. As a result, the field lacks empirical evidence on how BVI individuals judge which image filters suit different sensitive content and how these choices affect their comfort in sharing. Our research centers on two primary objectives: i) understanding BVI individuals’ comfort preferences regarding image filtering techniques across various image categories, and ii) exploring their comfort levels with sharing unfiltered and filtered images with friends, family, and strangers. To address these objectives, we conducted a user study with 20 BVI participants, who expressed their comfort levels while sharing images with different audiences, such as friends, family, and strangers. Additionally, participants were presented with images containing potentially sensitive or embarrassing content and were given options for filtering, including blurring, pixelation, masking, and content-based filling techniques. Results from the study showed that no single technique fits all situations - preferences varied by image and content type. Additionally, participants felt more comfortable sharing filtered images with audiences than non-filtered images. These findings emphasize the need for customizable, context-aware filtering tools that assist BVI users in making confident sharing decisions. Rather than proposing new algorithms, our contributions articulate how system decisions should adapt to BVI users’ context-specific privacy expectations, filling a critical gap between technical capability and end-user comfort. Our paper contributes to the field in the following ways: • An investigation of the preferences of BVI individuals regarding image filtering techniques across different types of sensitive content. • An examination how comfort levels in sharing images change for BVI users depending on the target audience before and after applying various image filters. • We present a set of design implications for image obfuscation techniques that are specifically tailored to the preferences of BVI individuals, supporting safer and more comfortable image sharing across different target audiences. 2. Related Work 2.1. Automated Assistive Systems Various camera-based assistive technologies, such as object identifiers, barcode readers, text readers, color readers, money readers, and crowd-sourced visual question-answering systems, have been developed to help BVIs with daily tasks like identifying objects, reading prescriptions, and answering questions ( 28 ; 34 ; 5 ; 10 ; 4 ) . The field of Human-Computer Interaction (HCI) has in recent years concentrated its research efforts on the intricate domain of private information within images ( 53 ; 40 ) , particularly those captured by individuals who are BVI ( 23 ; 19 ) . This line of investigation underscores the nuanced challenges confronted by the BVI community in harnessing visual assistive technologies effectively. One significant concern is the inadvertent capture of sensitive or confidential data within these images, leading to a notable hesitancy among BVI individuals when seeking visual assistance from sighted peers ( 48 ) . This hesitancy emanates from a fear of privacy breaches and the unintended exposure of personal information. Consequently, there exists a potential curtailment of BVI individuals’ engagement with the visual realm, a phenomenon that bears ramifications for