DescribePhotoA PictureEditor.com tool

WCAG 1.1.1 and GIGW, in plain terms

One success criterion governs nearly everything about images, it is two sentences long, and most of the rules attributed to it are not in it.

What the criterion actually says

Success criterion 1.1.1, Non-text Content, is a Level A requirement of WCAG 2.0, 2.1 and 2.2 alike, and its wording has not changed across any of them. Stripped to its spine: all non-text content presented to the user has a text alternative that serves the equivalent purpose, except in a set of listed situations.

Two words in that sentence do all the work. Equivalent — not identical, not exhaustive, not a transcription of everything visible, but whatever gets the same thing across. And purpose — the purpose the image is serving on this page, which is why the same photograph correctly gets different alt text in two different articles.

The listed exceptions are where the practical guidance lives. Controls and inputs need a name rather than a description. Time-based media, tests and sensory experiences get a short identifying label. CAPTCHA gets a description of its purpose plus an alternative form. And decoration, formatting and invisible content are to be implemented so that assistive technology can ignore them — which is the clause that makes an empty alt attribute a conformant answer rather than a shortcut.

How GIGW picks it up

GIGW 3.0, the Guidelines for Indian Government Websites published through the National Informatics Centre, sets WCAG 2.1 Level AA as the accessibility standard for government websites, portals and mobile applications. Level AA is cumulative: a site claiming it has claimed every Level A criterion underneath, 1.1.1 included. Behind the guideline sits the Rights of Persons with Disabilities Act, 2016, which is the statutory reason any of this has teeth.

What that means in an audit is less abstract than it sounds. A departmental site is typically assessed page by page against a checklist derived from those criteria, and the image findings are consistently among the most numerous — not because images are hard, but because there are hundreds of them and each one is a separate decision that somebody has to have made.

What an image finding usually is

A scanned circular published as a JPEG with no alternative at all. A departmental emblem whose alt text is the word “logo”. A banner carrying a scheme name and a deadline as pixels rather than as text. A PDF thumbnail linked with no accessible name. Each of those is a different criterion doing the work, and each has a different fix.

The one that needs a different fix

Notices published as images of text are the recurring case, and describing them better is not the answer. 1.4.5 Images of Text, at Level AA, asks that text be real text wherever presentation allows it. An image of a circular should become a page of text with the scan alongside it; writing a long alt attribute for it satisfies 1.1.1 and leaves 1.4.5 exactly where it was.

What a scanner can and cannot settle

Automated testing is genuinely good at a narrow set of these. A missing attribute, an empty accessible name on a linked image, alt text ending in a file extension, the same string repeated across every image in a gallery — all machine-detectable, all worth catching before a human ever looks.

What no scanner can settle is the criterion’s actual question. Whether this description serves an equivalent purpose depends on what the page is using the image for, and that is not in the markup. A page can pass every automated check on every image and still fail 1.1.1 comprehensively, and the reverse is true as well: a page full of flagged empty attributes may be doing exactly the right thing.

The checker on this site is in the first category and says so. It reads your string and reports what is wrong with the string. The judgement about equivalence is yours, and it is the part of the work that cannot be delegated to anything.

Questions people ask

Does 1.1.1 say how long alt text has to be?
No. It names no length, no format and no wording. It asks that non-text content have a text alternative serving an equivalent purpose, and it lists the situations where something shorter or nothing at all is the correct answer. Every character count you have been quoted is a convention someone built on top of it.
Is 1.1.1 Level A or Level AA?
Level A, which is the lowest of the three conformance levels and therefore the one that is hardest to argue your way out of. An organisation targeting AA has already accepted every Level A criterion; there is no conformance claim that includes AA and excludes this.
What does GIGW require, and who does it apply to?
GIGW 3.0 — the Guidelines for Indian Government Websites, published through the National Informatics Centre — sets WCAG 2.1 Level AA as the accessibility standard for government websites, portals and apps. Level AA includes every Level A criterion, so 1.1.1 comes with it. The Rights of Persons with Disabilities Act, 2016 sits behind that as the statutory obligation. Treat this page as an orientation rather than as legal advice; the published guideline document is the authority.
Will an automated audit tool tell me whether I have passed?
It will tell you where you have definitely failed. A missing alt attribute, alt text that is a file name, a linked image with no accessible name — those are machine-detectable and a scanner will find all of them. Whether a present, plausible description actually serves an equivalent purpose is not machine-detectable, and no scanner on the market can decide it.
Does text inside an image count as non-text content?
Yes, and two criteria apply. Under 1.1.1 the alternative has to carry the same information the image of text carries. Separately, 1.4.5 Images of Text at Level AA asks you not to use an image of text at all where real text would do — so an image of a paragraph is usually a problem to remove rather than a problem to describe.

Where to take this next

The other written pages