OpenAI debuts 'Screen-Reader Mode' for ChatGPT to improve spoken output and navigation

AI · 3 min read

Today OpenAI announced 'Screen-Reader Mode' for ChatGPT, a feature that reformats responses into readable blocks optimized for text-to-speech and screen readers. When enabled, the model outputs explicit headings, concise summaries, and semantic labels that reduce ambiguity in spoken interfaces and make long-form answers easier to skim via keyboard shortcuts.

The mode also adds inline hints designed for ARIA-aware parsers — for example, marking start and end of examples, flagging interactive suggestions, and breaking content into numbered steps when relevant. OpenAI says the redesign reduces cognitive load for listeners and improves navigation inside multi-part answers.

Accessibility advocates welcomed the move but urged OpenAI to publish detailed schema and to make the mode available through the API so developers can integrate it across assistive products. OpenAI confirmed API access is coming and pledged to work with disability organizations on further refinements.