New AI Accessibility Framework Maps Limits of Browser-Native Systems

In a new paper published on arXiv, researchers Rizwan Jahangir and Daisuke Ishii propose the Accessibility Capability Boundary (ACB), a formal framework for understanding how far AI-generated browser-native systems can push accessibility computing. Rather than treating accessibility as a binary compliance checkbox, the ACB models it as a dynamic, multidimensional capability space governed by variables like deployment latency, cognitive load, infrastructure dependency, offline persistence, interaction complexity, and adaptability. The authors argue that single-file HTML artifacts leveraging standard browser APIs can reduce deployment friction to near-zero and enable rapid, context-specific interface adaptation, effectively shifting the ACB outward.

The framework is grounded in two real-world exploratory prototypes. The first is an AI-generated browser-native accessibility interface deployed for a blind user in Nepal, demonstrating how such systems can work in low-infrastructure settings. The second is a fully functional, open-source webcam alignment assistant for visually impaired users. Through formal definitions, propositions, and a comparative evaluation matrix, the paper characterizes which regions of the accessibility capability space these systems can and cannot reach. It also identifies remaining computational, infrastructural, and verification constraints as hard boundaries. This work provides a theoretical foundation for understanding the scalable limits of autonomous accessibility computing and outlines a research agenda for future work.