OpenAI Adds Accessibility Mode to ChatGPT with Screen-Reader-Optimized Output

AI · 3 min read

The new Accessibility Mode in ChatGPT offers a streamlined response format by defaulting to plain-language summaries, shorter paragraphs, and explicit labels for actions and references. Users can enable an additional semantic layer that wraps responses in ARIA-like cues, such as "action:" "summary:" and "cite:" tags, which screen readers can parse more reliably than conversational prose.

OpenAI says the mode was developed with input from screen-reader users and accessibility researchers, focusing on reducing typical friction points: unclear pronoun references, long nested lists, and inconsistent headings. The model also supports an "explain layout" toggle that describes expected UI structures when users ask how to implement a component or layout in code.

Accessibility Mode is available across ChatGPT's web and mobile apps and is exposed in the API as an optional prompt parameter for enterprise customers building assistive workflows. OpenAI plans to continue iterating on the feature based on telemetry and community feedback, and it opened a public feedback channel specifically for assistive tech stakeholders.