Instagram automatically generates written descriptions of photos — called alternative text — so that screen reader apps used by blind or low-vision users can describe what is in an image out loud. This feature launched on Instagram specifically in November 2018.

The system uses a computer vision model (a type of machine learning that analyzes images) to identify objects, scenes, and activities in each photo uploaded to Instagram, producing descriptions such as "may contain: two people, smiling, outdoor, beach." These are read aloud by screen readers when a visually impaired user encounters the post. Creators can also write custom alternative text to override the AI-generated version. The feature was announced via the Instagram Blog and confirmed in the Instagram Help Center. The AI descriptions are often generic and can miss context a human description would include.