arXiv subjects

Robert Xiao

Publications and source records attributed to Robert Xiao.

17 recordsLinked to original sources

arXiv2026

Understanding Security and Privacy Perceptions of Content Creators Regarding AI Labels of AI-Generated Content

AI labels, typically implemented via underlying tracing mechanisms such as watermarks and metadata, are crucial for protecting Artificial Intelligence-Generated Content (AIGC) against security threats like disinformation and evasion. However, the perceived devaluation of AI-assisted work discourages creators from disclosing AI use, incentivizing efforts to bypass labeling and compromising downstream traceability. Yet, how AIGC creators perceive the security and privacy (S\&P) implications of these labels, and how their behaviors impact technical resilience remain underexplored. To this end, we conducted semi-structured interviews with 21 AIGC creators and measured images across 6 image generation platforms against 16 self-reported manipulation settings. Our findings reveal that creators conflate binary AI labels with granular traceability, and express strong fears of de-anonymization via platform identifiers. Driven by fears of algorithmic traffic suppression and reputational risks, they defensively removed digital traces. Through empirical tests, we show that targeted modifications like coarse quantization significantly degrade detection. AI detection capabilities are also inconsistent across platforms, and suffer from false positives even for human-authored images. Based on these insights, we advocate for workflow-resilient implicit AI labels that align technical guarantees with creators' incentives.

cs.HC↗

arXiv2026

Generative AI-Enabled Refund Fraud in Chinese E-Commerce: Investigation on Merchants and Platform Workers

E-commerce dispute resolution typically relies on the security assumption that digital evidence truthfully reflects physical reality. Generative AI (GenAI) invalidates this threat model, enabling attackers to fabricate hyper-realistic evidence of product defects at negligible cost. Through semi-structured interviews with merchants (N=17) and platform workers (N=13) in the Chinese e-commerce market, we characterize this shift toward GenAI-enabled scalable fabrication. We outline a taxonomy of four GenAI-enabled threat vectors across the transaction, dispute, logistics and communication phases, highlighting how attackers exploit GenAI to synthesize physically plausible product defects at scale. To mitigate these threats, platforms and merchants are adapting verification strategies, relying on AI tools for automated screening and adversarial interrogation (e.g., requesting multi-angle videos) to increase attack complexity. However, we find several challenges that hinder the adoption of these defenses, including implementation hurdles like structural platform constraints and fundamental limitations regarding the technical sophistication of GenAI. We conclude by outlining design implications for privacy-preserving cross-platform fraud databases, and traceability mechanisms such as embedding verifiable material anchors into the product.

cs.CR↗

arXiv2026

Investigating Novice Researchers' Perceptions of Research Privacy Within LLM-Assisted Workflows

Large Language Model (LLMs)-assisted scholarly workflows introduce critical privacy and intellectual property risks. As a uniquely vulnerable cohort driven by publication pressure and a lack of institutional support, novice researchers rely heavily on public LLMs, compelling them to navigate high-stakes privacy-publication trade-offs. To investigate these concerns, we conducted semi-structured interviews with 44 researchers across diverse disciplines. Our findings reveal that the fear of idea leakage paradoxically accelerates, rather than deters, reliance on LLMs, as researchers utilize them to expedite publication. They also held misconceptions that their ideas lacked the unique value to attract targeted attacks, and that their inputs would be safely diluted within massive datasets, preventing reconstruction. From interviews, we identified five types of mitigations including input fragmentation and adversarial probing, though we found that participants largely perceived these measures as ineffective. We outline implications including implementing institution-level sandboxed isolation, scenario-based privacy pedagogy, and verifiable data-deletion audits for transparency.

cs.HC↗

arXiv2026

"It Felt a Bit Eerie": Exploring Humanlike Interactions During Collaborative Writing with an Artificial Agent

While human-AI collaboration systems have increasingly been built to increase efficiency or support creativity, little work has examined how the design of interactions shapes the social connection between human and artificial agent. We examine how the temporal and visual dimensions of collaboration shape the experience of a writing task. Specifically, we built three variants of an AI-assisted text editor along a spectrum of simulated humanlike interaction (synchronous and with a cursor) to machinelike interaction (asynchronous and without a cursor), and conducted a comparative user study (n=48). Our exploratory findings suggest that synchronous suggestions increased efficiency but led to contextual misalignment, while a visual cursor increased intent understanding but evoked feelings of surveillance. Taken together, humanlike design of artificial agents can create positive social expectations but also elicit social costs, especially without the alignment present in human-human collaboration. We extend our findings into design implications and ethical considerations when building human-AI collaboration systems.

cs.HC↗

arXiv2026

"I Don't Have Faith in the Developers to Use My Feedback": Understanding Player Values and Expectancy for Reporting Systems in Video Games

Reporting systems in multiplayer video games allow players to express their dissatisfaction with others and combat in-game toxicity. In this work, we examined the act of reporting through the lens of expectancy-value theory. Using a distributed survey (n = 98) and follow-up interviews (n = 19), we explored the value players place on reporting, their desired outcomes, and their expectations that these outcomes will be achieved. Our findings revealed that reporting is motivated by both altruistic and retributive factors, with players seeking short-term revenge while also looking to foster an improved long-term community. Yet, players felt that reporting may not always meet these goals, with belief in the system being mediated by factors such as developer reputation, reporting transparency, and alignment with the community. By understanding the value and expectancy of reporting systems, we discuss their implications on broader digital moderation and consider current and potential future designs of reporting systems.

cs.HC↗

arXiv2026

Exploring Instant Photography using Generative AI: A Design Probe with the UnReality Camera

Generative AI has increasingly been used for artistic creation, but little work has explored how it shapes the experiential meaning of practice. We consider how generative AI might transform the embodied and tangible process of instant photography through the UnReality Camera, an AI-mediated instant camera. The UnReality Camera prints a photo of the environment augmented by a user's spoken words as generative input. In a design probe, we explored how generative AI shapes people's perceptions of both photographic output and the broader photographic process. Although users valued artistic control, they also appreciated the creativity afforded by stochastic unpredictability. The waiting period for an unpredictable output elicited anticipatory suspense, and the camera's physical form evoked ownership and connection despite artificial generation. We discuss how people make sense of instant photography's experiential qualities when generative AI is embedded, and how their opposing affordances reshape interpretations of each other's experiential meaning.

cs.HC↗

arXiv2026

Roomify: Spatially-Grounded Style Transformation for Immersive Virtual Environments

We present Roomify, a spatially-grounded transformation system that generates themed virtual environments anchored to users' physical rooms while maintaining spatial structure and functional semantics. Current VR approaches face a fundamental trade-off: full immersion sacrifices spatial awareness, while passthrough solutions break presence. Roomify addresses this through spatially-grounded transformation - treating physical spaces as "spatial containers" that preserve key functional and geometric properties of furniture while enabling radical stylistic changes. Our pipeline combines in-situ 3D scene understanding, AI-driven spatial reasoning, and style-aware generation to create personalized virtual environments grounded in physical reality. We introduce a cross-reality authoring tool enabling fine-grained user control through MR editing and VR preview workflows. Two user studies validate our approach: one with 18 VR users demonstrates a 63% improvement in presence over passthrough and 26% over fully virtual baselines while maintaining spatial awareness; another with 8 design professionals confirms the system's creative expressiveness (scene quality: 5.95/7; creativity support: 6.08/7) and professional workflow value across diverse environments.

cs.HC↗

arXiv2026

Reflective Motion and a Physical Canvas: Exploring Embodied Journaling in Virtual Reality

In traditional journaling practices, authors express and process their thoughts by writing them down. We propose a somaesthetic-inspired alternative that uses the human body, rather than written words, as the medium of expression. We coin this embodied journaling, as people's isolated body movements and spoken words become the canvas of reflection. We implemented embodied journaling in virtual reality and conducted a within-subject user study (n=20) to explore the emergent behaviours from the process and to compare its expressive and reflective qualities to those of written journaling. When writing-based norms and affordances were absent, we found that participants defaulted towards unfiltered emotional expression, often forgoing words altogether. Rather, subconscious body motion and paralinguistic acoustic qualities unveiled deeper, sometimes hidden feelings, prompting reflection that happens after emotional expression rather than during it. We discuss both the capabilities and pitfalls of embodied journaling, ultimately challenging the idea that reflection culminates in linguistic reasoning.

cs.HC↗

arXiv2026

The Words That Can't Be Shared: Exploring the Design of Unsent Messages

People often have things they want to say but hold back in conversations, fearing vulnerability or social consequences. Online, this restraint can take a distinctive form: even when such thoughts are written out - in moments of anger, guilt, or longing - people may choose to withhold them, leaving them unsent. This process is underexamined; we investigate the experience of writing such messages within people's digital communications. We find that unsent messages become expressive containers for suppressed feelings, where the act of writing creates a pause for reflection on the relationship and oneself. Building on these insights, we probe into how the design of the writing platforms of unsent messages affects people's experiences and motivations. Speculating with participants on nine evocative variants of a note-taking platform, we highlight how design shapes the emotional, temporal, and ritualistic qualities of unsent messages, revealing subtle tensions between people's social desires and communicative actions.

cs.HC↗

arXiv2026

Dissolving a Digital Relationship: A Critical Examination of Digital Severance Behaviours in Close Relationships

Fulfilling social connections are crucial for human well-being and belonging, but not all relationships last forever. As interactions increasingly move online, the act of digitally severing a relationship - e.g. through blocking or unfriending - has become progressively more common as well. This study considers actions of "digital severance" through interviews with 30 participants with experience as the initiator and/or recipient of such situations. Through a critical interpretative lens, we explore how people perceive and interpret their severance experience and how the online setting of social media shapes these dynamics. We develop themes that position digital severance as being intertwined with power and control, and we highlight (im)balances between an individual's desires that can lead to feelings of disempowerment and ambiguous loss for both parties. We discuss the implications of our research, outlining three key tensions and four open questions regarding digital relationships, meaning-making, and design outcomes for future exploration.

cs.HC↗

arXiv2025

Entertainers Between Real and Virtual -- Investigating Viewer Interaction, Engagement, and Relationships with Avatarized Virtual Livestreamers

Virtual YouTubers (VTubers) are avatar-based livestreamers that are voiced and played by human actors. VTubers have been popular in East Asia for years and have more recently seen widespread international growth. Despite their emergent popularity, research has been scarce into the interactions and relationships that exist between avatarized VTubers and their viewers, particularly in contrast to non-avatarized streamers. To address this gap, we performed in-depth interviews with self-reported VTuber viewers (n=21). Our findings first reveal that the avatarized nature of VTubers fosters new forms of theatrical engagement, as factors of the virtual blend with the real to create a mixture of fantasy and realism in possible livestream interactions. Avatarization furthermore results in a unique audience perception regarding the identity of VTubers - an identity which comprises a dynamic, distinct mix of the real human (the voice actor/actress) and the virtual character. Our findings suggest that each of these dual identities both individually and symbiotically affect viewer interactions and relationships with VTubers. Whereas the performer's identity mediates social factors such as intimacy, relatability, and authenticity, the virtual character's identity offers feelings of escapism, novelty in interactions, and a sense of continuity beyond the livestream. We situate our findings within existing livestreaming literature to highlight how avatarization drives unique, character-based interactions as well as reshapes the motivations and relationships that viewers form with livestreamers. Finally, we provide suggestions and recommendations for areas of future exploration to address the challenges involved in present livestreamed avatarized entertainment.

cs.HC↗

arXiv2025

VIBES: Exploring Viewer Spatial Interactions as Direct Input for Livestreamed Content

Livestreaming has rapidly become a popular online pastime, with real-time interaction between streamer and viewer being a key motivating feature. However, viewers have traditionally had limited opportunity to directly influence the streamed content; even when such interactions are possible, it has been reliant on text-based chat. We investigate the potential of spatial interaction on the livestreamed video content as a form of direct, real-time input for livestreamed applications. We developed VIBES, a flexible digital system that registers viewers' mouse interactions on the streamed video, i.e., clicks or movements, and transmits it directly into the streamed application. We used VIBES as a technology probe; first designing possible demonstrative interactions and using these interactions to explore streamers' perception of viewer influence and possible challenges and opportunities. We then deployed applications built using VIBES in two livestreams to explore its effects on audience engagement and investigate their relationships with the stream, the streamer, and fellow audience members. The use of spatial interactions enhances engagement and participation and opens up new avenues for both streamer-viewer and viewer-viewer participation. We contextualize our findings around a broader understanding of motivations and engagement in livestreaming, and we propose design guidelines and extensions for future research.

cs.HC↗

arXiv2025

HaloTouch: Using IR Multi-path Interference to Support Touch Interactions With General Surfaces

Sensing touch on arbitrary surfaces has long been a goal of ubiquitous computing, but often requires instrumenting the surface. Depth camera-based systems have emerged as a promising solution for minimizing instrumentation, but at the cost of high touch-down detection error rates, high touch latency, and high minimum hover distance, limiting them to basic tasks. We developed HaloTouch, a vision-based system which exploits a multipath interference effect from an off-the-shelf time-of-flight depth camera to enable fast, accurate touch interactions on general surfaces. HaloTouch achieves a 99.2% touch-down detection accuracy across various materials, with a motion-to-photon latency of 150 ms. With a brief (20s) user-specific calibration, HaloTouch supports millimeter-accurate hover sensing as well as continuous pressure sensing. We conducted a user study with 12 participants, including a typing task demonstrating text input at 26.3 AWPM. HaloTouch shows promise for more robust, dynamic touch interactions without instrumenting surfaces or adding hardware to users.

cs.HC↗

arXiv2024

VirtualNexus: Enhancing 360-Degree Video AR/VR Collaboration with Environment Cutouts and Virtual Replicas

Asymmetric AR/VR collaboration systems bring a remote VR user to a local AR user's physical environment, allowing them to communicate and work within a shared virtual/physical space. Such systems often display the remote environment through 3D reconstructions or 360-degree videos. While 360-degree cameras stream an environment in higher quality, they lack spatial information, making them less interactable. We present VirtualNexus, an AR/VR collaboration system that enhances 360-degree video AR/VR collaboration with environment cutouts and virtual replicas. VR users can define cutouts of the remote environment to interact with as a world-in-miniature, and their interactions are synchronized to the local AR perspective. Furthermore, AR users can rapidly scan and share 3D virtual replicas of physical objects using neural rendering. We demonstrated our system's utility through 3 example applications and evaluated our system in a dyadic usability test. VirtualNexus extends the interaction space of 360-degree telepresence systems, offering improved physical presence, versatility, and clarity in interactions.

cs.HC↗

arXiv2024

Streamlining Image Editing with Layered Diffusion Brushes

Denoising diffusion models have emerged as powerful tools for image manipulation, yet interactive, localized editing workflows remain underdeveloped. We introduce Layered Diffusion Brushes (LDB), a novel training-free framework that enables interactive, layer-based editing using standard diffusion models. LDB defines each "layer" as a self-contained set of parameters guiding the generative process, enabling independent, non-destructive, and fine-grained prompt-guided edits, even in overlapping regions. LDB leverages a unique intermediate latent caching approach to reduce each edit to only a few denoising steps, achieving 140~ms per edit on consumer GPUs. An editor implementing LDB, incorporating familiar layer concepts, was evaluated via user study and quantitative metrics. Results demonstrate LDB's superior speed alongside comparable or improved image quality, background preservation, and edit fidelity relative to state-of-the-art methods across various sequential image manipulation tasks. The findings highlight LDB's ability to significantly enhance creative workflows by providing an intuitive and efficient approach to diffusion-based image editing and its potential for expansion into related subdomains, such as video editing.

cs.CV↗

arXiv2023

Diffusion Brush: A Latent Diffusion Model-based Editing Tool for AI-generated Images

Text-to-image generative models have made remarkable advancements in generating high-quality images. However, generated images often contain undesirable artifacts or other errors due to model limitations. Existing techniques to fine-tune generated images are time-consuming (manual editing), produce poorly-integrated results (inpainting), or result in unexpected changes across the entire image (variation selection and prompt fine-tuning). In this work, we present Diffusion Brush, a Latent Diffusion Model-based (LDM) tool to efficiently fine-tune desired regions within an AI-synthesized image. Our method introduces new random noise patterns at targeted regions during the reverse diffusion process, enabling the model to efficiently make changes to the specified regions while preserving the original context for the rest of the image. We evaluate our method's usability and effectiveness through a user study with artists, comparing our technique against other state-of-the-art image inpainting techniques and editing software for fine-tuning AI-generated imagery.

cs.CV↗

arXiv2023

AutoDepthNet: High Frame Rate Depth Map Reconstruction using Commodity Depth and RGB Cameras

Depth cameras have found applications in diverse fields, such as computer vision, artificial intelligence, and video gaming. However, the high latency and low frame rate of existing commodity depth cameras impose limitations on their applications. We propose a fast and accurate depth map reconstruction technique to reduce latency and increase the frame rate in depth cameras. Our approach uses only a commodity depth camera and color camera in a hybrid camera setup; our prototype is implemented using a Kinect Azure depth camera at 30 fps and a high-speed RGB iPhone 11 Pro camera captured at 240 fps. The proposed network, AutoDepthNet, is an encoder-decoder model that captures frames from the high-speed RGB camera and combines them with previous depth frames to reconstruct a stream of high frame rate depth maps. On GPU, with a 480 x 270 output resolution, our system achieves an inference time of 8 ms, enabling real-time use at up to 200 fps with parallel processing. AutoDepthNet can estimate depth values with an average RMS error of 0.076, a 44.5% improvement compared to an optical flow-based comparison method. Our method can also improve depth map quality by estimating depth values for missing and invalidated pixels. The proposed method can be easily applied to existing depth cameras and facilitates the use of depth cameras in applications that require high-speed depth estimation. We also showcase the effectiveness of the framework in upsampling different sparse datasets e.g. video object segmentation. As a demonstration of our method, we integrated our framework into existing body tracking systems and demonstrated the robustness of the proposed method in such applications.

cs.CV↗