Can AI Fix the Web? The Promises and Perils of "Automated" Accessibility

29 Sep 2026, 3:35 pm (UTC)
20 Minutes

AI coding assistants can produce front-end code with hallucinated ARIA labels and broken accessibility. Automated fixes may pass audits while still creating poor screen-reader experiences, because AI does not replace human judgment in inclusive design.

The session examines where generative AI fails in accessibility work and how teams can use it as a testing and ideation partner rather than an automated fix. It covers spotting false accessibility fixes, writing context-aware prompts for AI to act as an accessibility auditor, and setting team guardrails for AI use in front-end development.

Other Sessions