AIccessiblity Content Updater

🌟 CF Hackathon 2025 Project Results 🌟

The AIccessibility team set out to create a proof-of-concept WordPress plugin that could automatically mitigate accessibility issues on websites by running background checks. Their initial plan focused on using AI to recognize text in images (common on e-commerce sites), identify missing “skip to content” links, and fix incorrectly formatted buttons by adding appropriate ARIA tags.

After consulting with accessibility experts at the hackathon, the team pivoted their approach to prioritize HTML5 tags over ARIA attributes whenever possible. They also shifted their “magic-wand” fix-everything concept to a more educational approach that highlights potential issues while teaching proper accessibility practices.

The team adapted to their technical strengths by developing a Linux command executable from PHP that provides machine-readable JSON output. Their image recognition system successfully extracted site context and matched it with machine vision results, adding a “relevancy” flag to determine whether to modify image alt tags or apply aria-hidden attributes based on a defined threshold.

Though the “Skip to Content” feature was only partially implemented due to challenges with whole-page screenshot processing, the team added an unplanned enhancement to WordPress’s media library that can generate alt tags for images based on website context. Despite remaining challenges with e-commerce and learning platforms, the project successfully demonstrated the feasibility of on-the-fly accessibility improvements, with team members committing to continue development in the coming months.

Pitch Interview Video

Project Description

A WordPress plugin will enhance website accessibility. It will automatically add skip-to-content links, improve alt text for images using AI, and add ARIA attributes to buttons. The plugin will operate behind the scenes, processing HTML during page rendering. The initial focus is on basic WordPress sites; WooCommerce support is a future consideration. The plugin’s functionality will rely on an existing AI service via API.

Target Audience

The plugin targets WordPress website owners. A subscription to an AI service will be required for the Proof Of Concept. 

Skills needed:

• Prompt engineering
• DOM manipulation (language agnostic)
• Caching system development (language agnostic)
• WordPress hook implementation (PHP)
• Accessibility testing (screen reader experience helpful)
• Image analysis & detection (screenshot engine)

AI expertise is not required; an existing AI service will be used via API. Open-source LLMs are not considered for this hackathon due to time constraints and the complexity of image analysis.

Hackathon Goals

Develop a proof-of-concept plugin demonstrating improved website accessibility. Three key features will be implemented: skip-to-content, improved alt text, and ARIA attributes. Pre- and post-implementation metrics will be tracked to assess effectiveness.