WCAG

WCAG is the W3C’s standard for making web content accessible to disabled people, and the reference point most accessibility laws and policies use. It sets out what content and interfaces need to achieve rather than how to build them, so it applies to text, images, video, forms and documents as well as code.

It’s built in layers. Four principles say content must be perceivable, operable, understandable and robust. Under them sit guidelines, 13 in WCAG 2.2, and under those, testable success criteria, which are what an audit checks. Each criterion has a level: A for the most basic requirements, AA for the usual target and AAA for the most demanding. Most laws and procurement rules point to level AA of WCAG 2.1 or 2.2, so name the version and level when setting a target.

WCAG 1.0 was published in 1999. WCAG 2.0 followed in 2008 with the technology-neutral, testable approach still used today, and versions 2.1 (2018) and 2.2 (2023) added criteria for mobile, low vision and cognitive needs. WCAG 3, a broader rewrite with a different structure, is still a working draft and isn’t expected to replace WCAG 2 for some years.

Why it matters for content

WCAG gives teams a shared standard rather than leaving accessibility to taste or intention. It supplies testable criteria, but conformance doesn’t establish that every disabled person can use the experience, so involve people and look at needs beyond the checklist too.

Many criteria are met or failed by editorial choices before any developer touches the code. Writers and editors own text alternatives for images (1.1.1), headings and structure that reflect the content (1.3.1 and 2.4.6), page titles (2.4.2), link text that makes sense in context (2.4.4), labels and instructions on forms (3.3.2), error messages that say what went wrong and how to fix it (3.3.1 and 3.3.3), not relying on colour alone (1.4.1) and captions and transcripts for audio and video. At AAA, reading level (3.1.5) asks for supplementary content when text needs more than lower secondary education to understand.

Conformance covers the content and service as a whole, including editorial choices, so a site can fall out of conformance because of what’s published on it long after launch.

How AI changes the picture

AI makes the standard more important, because content is now generated, transformed and summarised in more places. WCAG gives teams a way to judge whether those outputs are usable.

AI can help with some checks and first-pass fixes, such as drafting alt text or flagging vague links. It can also create failures at scale: bland link text, alt text that describes an image without conveying its purpose, summaries that drop key steps or reformatted content that loses its heading structure. The existence of a machine doesn’t lower the standard.

Automated testing catches only some WCAG failures. Criteria that depend on meaning, such as whether alt text is appropriate or a heading describes its section, still need a person to judge.