Tag
The 2026 WebAIM Million report found 56.1 accessibility errors per home page on average, a 10.1% increase from 2025, with home page complexity rising 22.5% in one year.
This paper presents a method for training a lightweight vision-language model via reinforcement learning to detect violations of 19 UI quality principles in generated web pages, achieving 84% micro-F1. The model serves as a critic for auditing generated interfaces and providing design-aware feedback.
A study by AccessLumens found that 42% of US city websites still have critical accessibility barriers after the ADA Title II deadline, with an average score of 93 out of 100 but significant failures in key criteria. The choice of CMS vendor strongly correlates with accessibility risk.
The article criticizes the misuse of ARIA and the ARIA Authoring Practices Guide (APG), warning against using LLM agents to generate code from the APG, and emphasizes that native HTML should be preferred over ARIA for accessibility.
This article explains why using aria-label on generic HTML elements like divs or spans violates ARIA spec and leads to inconsistent screen reader behavior, with browser testing results.