How to Use AI to Write Alt Text for WordPress Images

Most image libraries in WordPress are full of gaps. A product photo gets uploaded with the filename still reading IMG_4213.jpg, the alt text field stays empty, and nobody notices until an accessibility audit or a lost bit of image-search traffic points it out. Writing a good description for every photo, screenshot, and graphic on a site is tedious enough that it’s one of the first things skipped under deadline pressure.

AI tools have made this far less painful. Feed an image to a vision-capable model and it will draft a reasonable description in seconds — the catch is knowing when to trust that draft and when to rewrite it. In my experience, AI-generated alt text is a strong starting point but rarely publish-ready as-is.

The “125-Character Limit” Isn’t a Real Rule

Almost every alt text guide repeats some version of a 100, 125, or 150-character limit as if it were a technical requirement. It isn’t. WCAG — the actual international accessibility standard alt text is supposed to satisfy — sets no character limit anywhere. The 125-character figure traces back to older versions of JAWS, the most widely used Windows screen reader, which used to truncate long alt attributes at around that length before moving on. That old technical quirk got repeated across SEO and accessibility guides until it hardened into a rule nobody questions. Modern screen readers don’t cut text off — JAWS today splits long alternative text across multiple announcements instead of truncating it, and will read the whole thing either way.

That doesn’t mean write long descriptions by default — concise is still the right instinct, because a screen reader user has to sit through every word before moving to the next element on the page, and a bloated description is genuinely annoying to listen to. Treat the 125-character range as a useful style habit for keeping AI drafts honest and easy for a human editor to scan, not a hard technical ceiling. The actual test is whether the description conveys what a sighted visitor gets at a glance — sometimes that’s eight words, occasionally it genuinely needs more.

Why This Matters for Two Different Audiences

Alt text serves two audiences at once. Screen reader users hear it read aloud in place of the image, so it needs to convey the same information a sighted visitor gets at a glance. Search engines also read it to understand what an image shows, since they can’t “see” pixels the way a person does — it’s one of the few direct signals you can give about image content, alongside the filename and surrounding text covered in optimising images for SEO.

Skipping it isn’t a small oversight. Missing alt text is one of the most common findings in any accessibility review, and it sits alongside the other basics covered in making a WordPress website accessible — colour contrast, keyboard navigation, and readable link text. Getting the images right is usually the fastest of those to fix, provided you’re not starting from a blank field on every upload.

The Workflow

Step 1: Audit what’s missing

Before generating anything, find out how bad the gap actually is. In the WordPress media library, switch to list view and scan the Alt Text column — it’s blank for any image that’s never had one written. On a large site, a free plugin like Media Library Assistant can filter specifically for images missing alt text, which saves scrolling through hundreds of entries by hand.

Step 2: Pick an AI tool

Two approaches work well. For a handful of images, upload each one to a chatbot with vision support and ask it to describe what’s shown in one plain sentence, without marketing language. For a whole media library, an AI-powered alt text plugin can queue every image and generate descriptions in bulk using the same kind of model behind the scenes.

Step 3: Generate a first draft

Ask for a factual description of what’s in the frame — subject, action, and any text visible in the image — rather than a caption or a sales pitch. A prompt like “describe this image in one sentence for a screen reader user, focusing on what’s visually shown” produces far more usable results than a generic “write alt text for this.”

Step 4: Edit every description

Read each draft against the image and the page it sits on. AI models are good at naming objects and can’t always tell what’s relevant to your specific page — a product photo might get described purely by colour and shape when the detail that actually matters is the brand or model name. Trim anything generic (“image of,” “picture showing”) and add the one or two details a reader actually needs, per Google’s guidance on writing helpful image descriptions. This editing pass is the step that’s easiest to skip when you’re clearing a large backlog in one sitting, but it’s also the one that determines whether the whole exercise was worth doing.

Step 5: Add it in WordPress

Open the image in the media library, or select it from within the block editor, and paste the finished description into the Alternative Text field in the right-hand panel. Leave the caption field for anything meant to display on the page — alt text and captions serve different purposes and shouldn’t just repeat each other.

Practical Tips

• Aim for roughly 125 characters as a default working target, not a hard cutoff — go shorter when the image is simple, and don’t panic if a genuinely detailed image needs an extra sentence.
• Purely decorative images — a divider line, a background texture — don’t need a description at all. Leave the alt text field empty rather than writing “decorative image,” which screen readers will still announce unnecessarily.
• For images that already contain the exact same information as the surrounding paragraph, a short alt text is fine — you don’t need to repeat the whole caption inside it.
• If a chatbot’s response comes back as a full paragraph, don’t paste it in as-is — cut it down to the sentence or two that would actually help someone who can’t see the image.

Common Mistakes

• Publishing the AI draft unedited. Vision models occasionally misread what’s in an image, especially with low-contrast screenshots or busy product photos — always check the description matches reality before saving it.
• Treating 125 characters as a hard technical limit and truncating a description mid-sentence to hit it, when the actual constraint is clarity, not character count.
• Stuffing target keywords into every alt text field. It reads as spam to anyone using a screen reader and doesn’t help rankings the way a genuinely descriptive sentence does.
• Writing the same generic description for every image in a gallery, which defeats the purpose for screen reader users trying to tell the photos apart.

When AI-Assisted Alt Text Is Worth It

AI-assisted alt text makes sense once a library has more than a few dozen images, or when you’re clearing a long-standing backlog — writing hundreds of descriptions from scratch by hand simply doesn’t happen in practice. For a small site with a handful of photos, writing each description manually takes about as long as reviewing an AI draft, so there’s no real time saved and it’s just as easy to skip the tool entirely.

AI can turn alt text from a chore into a five-second review step, but only if every generated description gets checked against the actual image before it’s saved. Start with the images missing alt text entirely, since that’s where the accessibility and SEO gap is largest, and don’t let a decades-old JAWS quirk dictate how you write for readers today.