Alt text, or alternative text, is a written description placed in an image's HTML alt attribute that serves as a replacement when the image cannot be seen. It does several jobs at once: screen readers read it aloud to people with visual impairments, search engines use it to understand and index the image, AI systems read it because they cannot see the image itself, and browsers display it when an image fails to load. Providing alt text for meaningful images is a Level A requirement of WCAG 2.2 under success criterion 1.1.1, which makes it a legal baseline, not an optional extra.

How long should alt text be?

Aim for 125 characters or fewer. This is not a strict WCAG rule, but several screen readers, including JAWS, truncate or change behaviour around that length, and overly long descriptions overwhelm the people relying on them. Write a specific, concise description of what the image shows and why it matters in context — 'black leather backpack on a wooden table' rather than 'bag' or a paragraph. For complex images like charts, keep the alt text to a short summary and provide the full detail in text nearby on the page.

What is the difference between empty alt and missing alt?

It is one of the most important distinctions in image accessibility. An empty alt attribute, written as alt="", tells a screen reader to skip the image silently, which is exactly what you want for decorative images. A missing alt attribute is different and worse: when no alt is present, many screen readers fall back to reading the image's file name aloud, so a user hears something like 'hero-banner-final-v2-dot-jpg', which is meaningless noise. Decorative images therefore need an explicit empty alt, never an omitted attribute.

Should alt text start with 'image of'?

No. Screen readers already announce that an element is an image before they read the alt text, so beginning with 'image of' or 'picture of' is redundant and wastes the listener's time — simply describe what the image shows. There is one exception: when the medium itself carries meaning, naming it adds useful context, as in 'Oil painting of sunflowers by Van Gogh' or 'Screenshot of the checkout page'. Otherwise, lead straight with the description.

Can AI write alt text for me?

AI vision models can produce a useful first draft, especially for simple informative images, but they should never be the final word. They frequently miss the context that decides good alt text, write generic descriptions, struggle with functional and complex images, and sometimes describe the image inaccurately. The reliable workflow is to use AI as a starting point and have a person edit for accuracy and context — because the right alt text depends on why the image is on that particular page, which only the people building it fully know.

Alt text and accessible images: one tag that serves accessibility, SEO and AI

June 26, 2026 · 11 min read · Web Involved

What is alt text, and how do you write it well?

Alt text is a short description placed in an image’s HTML alt attribute, and it is the rare detail that serves four audiences with one line: it is read aloud to screen-reader users, read by search engines indexing your images, read by AI answer engines that cannot see images at all, and shown to anyone whose image fails to load. Providing it for meaningful images is a Level A requirement of WCAG — the legal minimum — and missing alt text is the single most common accessibility failure on the web. Writing it well is a decision tree rather than a formula: if an image is decorative, give it an empty alt="" (never an omitted attribute); if it is informative, describe what it means in the context of the page; if it is functional, describe the action or destination rather than the picture; and if it is complex, write a short summary and put the full detail in text nearby. Keep it specific and concise, around 125 characters, skip “image of,” and let the context of the page decide what matters.

In short

Alt text is one HTML attribute that does four jobs at once: it's read aloud to screen-reader users, read by search engines, read by AI engines that can't see images, and shown when an image fails to load.
It's the law and the most common failure: WCAG 1.1.1 (Non-text Content) is Level A, yet missing alt text is the #1 web accessibility violation, found on 54.5% of home pages — and the most-cited issue in accessibility lawsuits.
The craft is a decision tree: decorative → empty alt='' (never a missing attribute); informative → describe the meaning in context; functional → describe the action; complex → short summary plus full detail nearby.
Write specific and concise (~125 characters), skip 'image of,' don't keyword-stuff, and let context decide. AI can draft simple alt text but misses context and function — it's a starting point, not the finish.

What is alt text — and the four jobs it does

Alt text is alternative text: a written description added to an image’s HTML alt attribute that stands in for the image when it cannot be seen (RatedWithAI, 2026). Its value is that a single attribute serves several audiences at the same time. Screen readers read it aloud to people with visual impairments, search engines use it to understand the image’s content, and browsers display it when an image fails to load — three jobs from one line (ContentPen, 2026). In 2026 there is a fourth: AI systems like ChatGPT, Claude and Perplexity cannot see images, so they read the alt text, and rich descriptions make your images legible to the answer engines that increasingly mediate discovery (AltText.ai, 2026).

That combination is why alt text sits at the intersection of accessibility, usability and search. It adds information for blind and low-vision screen-reader users, but it also benefits people on poor connections, supports some cognitive disabilities, and improves search-engine understanding of your pages (Siteimprove, 2026). Few elements on a page repay a small effort across so many fronts, which is exactly why getting it right is worth a deliberate method rather than a guess.

Why it’s not optional: the law and the lawsuits

Alt text is a legal baseline, and the gap is enormous. Providing a text alternative for non-text content is WCAG 2.2 success criterion 1.1.1, set at Level A — the minimum level of accessibility compliance — so meaningful images without alt text fail the most basic conformance bar (AccessibilityChecker, 2026). Yet missing alt text is the single most common accessibility violation on the web, found on 54.5% of all home pages in WebAIM’s survey, and it is the most common WCAG failure cited in accessibility lawsuits (RatedWithAI, 2026).

The legal exposure is concrete and rising. The United States Department of Justice’s ADA Title II rule requires state and local governments serving populations of 50,000 or more to meet WCAG 2.1 Level AA by 24 April 2026, with federal penalties of up to $75,000 for a first violation and $150,000 for subsequent ones (RatedWithAI, 2026). The European Accessibility Act took effect in June 2025, Canada’s Accessible Canada Act sets deadlines for 2027 and 2028, and alt text requirements are foundational to every one of these international standards (RatedWithAI, 2026). The broader legal landscape is the subject of our pillar on whether web accessibility is the law — alt text is simply where most sites fail it first.

The decision tree: what kind of image is it?

Good alt text starts before you write a word, with one question. Ask whether the image adds information a visitor would miss if it were not there — the mental test is to imagine removing the image entirely, and if the page still makes complete sense, the image is decorative (ContentPen, 2026). From there, the W3C’s decision tree sorts every image into a small number of cases (W3C WAI, 2026).

The categories are: decorative images, which carry no information and should receive an empty alt; informative images such as photographs and simple diagrams, which need a brief description of the meaning they convey; functional images used inside links or buttons, which need the function described; complex images like charts and infographics, which need a short alt plus a longer description nearby; and images of text, where the text in the image should be placed in the alt (Alt Audit, 2026). Categorising first is what stops you writing a careful description for an image that should have been silent, or staying silent on one that carries the page’s whole point.

The cardinal mistake: empty alt vs missing alt

This is the distinction that trips up the most sites, and it matters more than it looks. A decorative image should have an explicit empty alt attribute, written alt="", which signals to a screen reader that the image should be skipped, preventing unnecessary noise (Alt Audit, 2026). The critical error is omitting the attribute entirely, because a missing alt is treated very differently from an empty one.

When a screen reader meets an image with no alt attribute at all, it commonly falls back to reading the file name aloud, so the user hears something like “hero-banner-final-v2-dot-jpg” — a genuine distraction with no meaning (Siteimprove, 2026). The rule that follows is simple: decorative images need an empty alt, not a deleted one. The same logic applies to redundancy — an icon sitting next to a text label that already says the same thing, or a logo whose link text already names the destination, should get an empty alt so the screen reader does not announce the same information twice (AllAccessible, 2026).

How to write good alt text

For images that do need a description, a few rules cover almost every case. Be specific rather than vague — “black leather backpack on a table with a green background” tells a screen-reader user something, while “bag” or “photo” tells them nothing (Verbit, 2026). Be concise: aim for around 125 characters, since several screen readers truncate or change behaviour beyond that length, and a wall of description is as unhelpful as none (RatedWithAI, 2026). And do not begin with “image of” or “picture of,” because the screen reader has already announced that it is an image, with one exception — when the medium itself adds meaning, as in “Oil painting of sunflowers” or “Screenshot of the checkout page” (RatedWithAI, 2026).

Two habits keep alt text honest. Let context decide, because the same image needs different alt text depending on where it sits and what the surrounding words already say (Harvard, 2026). And resist keyword stuffing: cramming search terms into the attribute makes it unreadable for screen-reader users and can draw a search penalty, so any keywords should appear only where they describe the image naturally (AccessibilityChecker, 2026). A caption, by contrast, is the visible text below an image that everyone sees; it can coexist with alt text but does not replace it, since the two serve different readers (ContentPen, 2026).

Functional and complex images

Two categories deserve their own rules because the obvious description is the wrong one. For a functional image — one used inside a link or button — describe the action or destination, not the picture: a logo linking home takes alt="Web Involved home", and a sharing icon takes alt="Share on Twitter", because what the user needs to know is what happens when they activate it (AllAccessible, 2026). Describing the logo’s appearance there would tell the user nothing about where the link goes.

Complex images need a two-layer approach. For a chart, infographic or dense diagram, write a brief alt text summarising the key takeaway, then provide the full data or a detailed description elsewhere on the page, linked with aria-describedby where appropriate — WCAG 1.1.1 effectively asks for both the short alternative and the detailed one (RatedWithAI, 2026). A useful pattern is an alt like “Bar chart: organic traffic rose 40% from Q1 to Q2 after a content refresh,” with the underlying numbers in a nearby table that every visitor can read (ContentPen, 2026). And where an image contains text that matters, that text belongs in the alt so screen-reader users get the same information everyone else does (Siteimprove, 2026).

The SEO and AI payoff

Beyond access, alt text earns its keep in discovery. Search engines use it to index and understand images, which improves image-search visibility and overall discoverability — one of the easiest shared wins for accessibility and SEO alike (Verbit, 2026). The example writes itself: alt="Bar chart showing a 40% increase in organic traffic from Q1 to Q2" serves accessibility, supports SEO, and provides a broken-image fallback in a single line (ContentPen, 2026).

The AI dimension is newer and sharper. Because models like ChatGPT, Claude and Perplexity cannot see images, the alt text is what they read — the difference between an engine seeing “polar-bear-snow.jpg” and seeing “a polar bear standing on ice” decides whether your image is understood and surfaced (AltText.ai, 2026). Descriptive alt text gets your images included when an AI assembles an answer or recommendation, which makes it part of the same machine-readability discipline our pillar on whether your website is readable by AI describes. The file half of image quality — formats and loading speed — is covered in our guide on optimizing images for the web; alt text is the meaning half.

Can AI write your alt text?

It can help, with a clear boundary. AI vision models have matured enough to generate contextually useful descriptions for many images, and tools built on them can draft alt text at a scale no human can match across a large catalogue (Alt Audit, 2026). For simple informative images, a model like a current vision system produces a reasonable first draft (ContentPen, 2026).

The boundary is that the draft is not the finish. AI-generated alt text should always be reviewed by a person, because it often misses context, writes generic descriptions, and can make accuracy errors, and it struggles most with exactly the functional and complex images where the stakes are highest (RatedWithAI, 2026). The deeper reason is structural: the right alt text depends on why an image is on a particular page, and that intent lives with the people building the page, not in the pixels. It is the same limit that makes an accessibility overlay unable to fix this for you, the subject of our guide on why overlays are a legal trap — a script bolted on afterward cannot know what an image was meant to communicate.

How we do it

We treat the alt attribute as a required field, decided when an image is placed, not patched in later. Every meaningful image gets a description written for its context; every decorative one gets an explicit empty alt so screen readers pass over it cleanly; functional images describe their action; and charts carry a short summary with the full figures in real text nearby. Because that work happens at authoring time, the four audiences are served at once and none of them is an afterthought.

This is the clearest small example of the principle behind everything we build: get it right at the source and one decision serves a screen-reader user, a search engine, an AI answer engine, and a visitor on a slow connection — bolt it on later and you serve none of them well. Alt text is the operable, perceivable detail that our pillar on whether web accessibility is the law treats as a baseline, alongside the colour contrast and keyboard decisions that, like this one, are made while building or not at all.

Frequently asked

What is alt text?: Alt text, or alternative text, is a written description placed in an image's HTML alt attribute that serves as a replacement when the image cannot be seen. It does several jobs at once: screen readers read it aloud to people with visual impairments, search engines use it to understand and index the image, AI systems read it because they cannot see the image itself, and browsers display it when an image fails to load. Providing alt text for meaningful images is a Level A requirement of WCAG 2.2 under success criterion 1.1.1, which makes it a legal baseline, not an optional extra.
How long should alt text be?: Aim for 125 characters or fewer. This is not a strict WCAG rule, but several screen readers, including JAWS, truncate or change behaviour around that length, and overly long descriptions overwhelm the people relying on them. Write a specific, concise description of what the image shows and why it matters in context — 'black leather backpack on a wooden table' rather than 'bag' or a paragraph. For complex images like charts, keep the alt text to a short summary and provide the full detail in text nearby on the page.
What is the difference between empty alt and missing alt?: It is one of the most important distinctions in image accessibility. An empty alt attribute, written as alt="", tells a screen reader to skip the image silently, which is exactly what you want for decorative images. A missing alt attribute is different and worse: when no alt is present, many screen readers fall back to reading the image's file name aloud, so a user hears something like 'hero-banner-final-v2-dot-jpg', which is meaningless noise. Decorative images therefore need an explicit empty alt, never an omitted attribute.
Should alt text start with 'image of'?: No. Screen readers already announce that an element is an image before they read the alt text, so beginning with 'image of' or 'picture of' is redundant and wastes the listener's time — simply describe what the image shows. There is one exception: when the medium itself carries meaning, naming it adds useful context, as in 'Oil painting of sunflowers by Van Gogh' or 'Screenshot of the checkout page'. Otherwise, lead straight with the description.
Can AI write alt text for me?: AI vision models can produce a useful first draft, especially for simple informative images, but they should never be the final word. They frequently miss the context that decides good alt text, write generic descriptions, struggle with functional and complex images, and sometimes describe the image inaccurately. The reliable workflow is to use AI as a starting point and have a person edit for accuracy and context — because the right alt text depends on why the image is on that particular page, which only the people building it fully know.