Signals of Provenance: Practices & Challenges of Navigating Indicators in AI-Generated Media for Sighted and Blind Individuals

TL;DR

This work tackles the accessibility gap in AI provenance signaling for AI-generated media by comparing sighted and BLV users. It uses semi-structured interviews and interactive tasks across major platforms to reveal how users interpret and interact with existing indicators. The study uncovers four mental models, demonstrates that content-based cues often outweigh platform labels, and highlights significant accessibility barriers for BLV users. It proposes design and policy directions, including canonical disclosure schemas, API standardization, and governance registries, to make AI provenance signals trustworthy and accessible for diverse users.

Abstract

AI-Generated (AIG) content has become increasingly widespread by recent advances in generative models and the easy-to-use tools that have significantly lowered the technical barriers for producing highly realistic audio, images, and videos through simple natural language prompts. In response, platforms are adopting provable provenance with platforms recommending AIG to be self-disclosed and signaled to users. However, these indicators may be often missed, especially when they rely solely on visual cues and make them ineffective to users with different sensory abilities. To address the gap, we conducted semi-structured interviews (N=28) with 15 sighted and 13 BLV participants to examine their interaction with AIG content through self-disclosed AI indicators. Our findings reveal diverse mental models and practices, highlighting different strengths and weaknesses of content-based (e.g., title, description) and menu-aided (e.g., AI labels) indicators. While sighted participants leveraged visual and audio cues, BLV participants primarily relied on audio and existing assistive tools, limiting their ability to identify AIG. Across both groups, they frequently overlooked menu-aided indicators deployed by platforms and rather interacted with content-based indicators such as title and comments. We uncovered usability challenges stemming from inconsistent indicator placement, unclear metadata, and cognitive overload. These issues were especially critical for BLV individuals due to the insufficient accessibility of interface elements. We provide practical recommendations and design implications for future AIG indicators across several dimensions.

Figures (9)

• Figure 1: User Interface of YouTube and TikTok's AI info label. (a) single-line label "Altered or synthetic content" and hidden description "How this content was made" (b) single-line label "Creator labeled as AI-generated"
• Figure 2: User Interface of Meta's AI info label. "AI info" label is embedded in the post header, rotating with other elements like the location tag.
• Figure 3: Participants’ Mental Model for AI-generated Content - four dimensions derived from our qualitative results.
• Figure 4: Example screenshots of title and comments that participants accessed during interviews. (Left) The title is positioned very close to the content, making it easily accessible without the need to scroll. (Right) Comments offer nuanced insights, but they are not always clear or straightforward.
• Figure 5: Example screenshots of descriptions and hashtags that participants accessed during interviews. (Left) Descriptions are placed in highly visible areas but are sometimes overlooked because users focus on the main content. (Center, Right) Hashtags are easy to notice but can be unreliable, as they are sometimes used with unrelated tags to increase reach.