Detects alternative text that does not correctly describe the content of an image. Alternative text exists to provide a textual equivalent for users who cannot see visual content, whether due to visual impairments, slow internet connections, or disabled images. When alternative text is inaccurate or invalid, it becomes misleading rather than helpful, failing to convey the intended information and undermining accessibility for a significant portion of users.

Alternative text should clearly describe the content it represents. Identifies alternative text that is too vague or generic to be useful, such as “logo”, “image”, “picture”, “photo”, or “flag”. These terms fail to convey meaningful information and do not help users understand the content of the image.

Flags cases where the alternative text is written in a different language than the one specified in the containing element’s lang attribute or the document’s language. When there is a mismatch, screen readers may mispronounce the alternative text, making it difficult or impossible for users to understand.

Identifies symbols or special characters in alternative text that produce unwanted or confusing output when read aloud by screen readers. This includes repeating punctuation marks such as “—” or “***” and decorative separators like |, /, and –. Such characters create unnecessary audio clutter for screen reader users, distracting them from the main content. These symbols should not appear in alternative text unless they are part of the actual visible text in the image.

Detects charts, graphs, and other data visualizations that lack meaningful alternative text. Without a text description, users who cannot see the visual content have no way to access the data being presented.

Detects spelling and grammar mistakes in alternative text. Such errors can confuse screen reader users, reduce clarity, and make images harder to understand. Correct, clear alternative text ensures accurate interpretation for people relying on assistive technologies.

Identifies alternative text that has been used to insert brand names, marketing slogans, or other promotional content rather than providing a genuine description of the image. This gives users no meaningful visual information, making the alternative text useless for screen readers and reducing understanding for people who rely on them.

Identifies alternative text that appears to be a filename rather than a genuine description of the image. Using a filename as alternative text is a common malpractice in web development, often done to suppress accessibility warnings under the assumption that the filename is a good enough description. It never is. Filenames such as “IMG_2045.jpg” or “header-banner-v2.png” provide no useful information, and because they contain characters such as dashes and underscores as well as file extensions, they cause additional problems for screen reader users who must listen to them being read aloud.

Detects when text on a page is written in a different language than the one declared for the document, without the language change being explicitly indicated using the lang attribute. ACAI detects the language of each text block in the document body and compares it to the document’s declared language. Text blocks that differ from the declared language and are not wrapped in an element with the correct lang attribute are flagged as issues.

Detects when the declared language of a specific element does not match the actual language of the text within it. When an element uses the lang attribute, the specified language must accurately reflect the content. ACAI detects the language of each text block in the document body and compares it to the declared language of its containing element, flagging mismatches that could cause screen readers to mispronounce the content.

Detects alternative text that includes phrases such as “image of” or “picture showing”. These phrases are redundant because screen readers already announce images. Including them adds noise, slows navigation, and makes content less efficient for users relying on assistive technology.