Can AI crawlers read JavaScript sites? We read the docs

We read the crawler documentation of four AI platforms on 13 August 2026 and counted the named fetch paths. Eighteen. One is documented as executing JavaScript, one is documented as not supporting JavaScript-rendered pages, and the other sixteen say nothing about it either way.

That silence is the finding. It is not proof that those crawlers cannot render a page, and we are not claiming it is. It does mean that if your content only exists after JavaScript runs, you are betting on behaviour nobody has published.

Which AI crawlers say they execute JavaScript?

Google is the only one of the four that answers plainly. Its JavaScript SEO documentation states that Google Search runs JavaScript with an evergreen version of Chromium, and that a headless Chromium renders the page once resources allow.

• Google: documented, though the statement sits on its JavaScript page rather than on any crawler reference page
• OpenAI: four named agents, OAI-SearchBot, OAI-AdsBot, GPTBot and ChatGPT-User, with no rendering statement
• Perplexity: two named agents, PerplexityBot and Perplexity-User, with no rendering statement
• Anthropic: one documented fetch path, stated as not supporting JavaScript-rendered sites

The OpenAI crawlers page describes what each of its four agents is for, which robots.txt token controls it, and where to find its IP ranges. Rendering is not mentioned.

The same holds for the two agents on the Perplexity crawlers page. Both are described by purpose, user agent string and published IP ranges, and rendering does not come up.

Why a Chrome version in the user agent proves nothing

OAI-SearchBot advertises itself with a Chrome version inside its user agent string. So does Googlebot. A user agent string is a label a client chooses to send, not a capability statement, and reading one as evidence of rendering is the most common mistake we see on this question.

"Server-side or pre-rendering is still a great idea because it makes your website faster for users and crawlers, and not all bots can run JavaScript."

That is Google, in its own JavaScript documentation, telling you not to assume. If the company that does render is warning you that others may not, the safe read is to serve the content in HTML. We covered the related control gap in our note on what LLM SEO actually rests on.

What Google actually documents about rendering

Google publishes 11 common crawlers on its crawler reference page, and a separate technical properties page covering protocols, encodings, file size limits, crawl rate and HTTP caching. Neither page states rendering behaviour for any agent. You have to leave the crawler documentation entirely to find it.

The Googlebot page adds a limit worth knowing. Googlebot crawls the first 2MB of a supported file type, and each resource referenced in the HTML, including your JavaScript, is fetched separately and bound by that same limit. A bundle above it gets truncated.

What does a non-rendering fetch actually see?

We have measured this. In August 2026 we fetched the homepages of 30 Tamil Nadu engineering colleges as raw HTML with JavaScript disabled, which is close to what a fetch path that does not render would receive. Five of the 30 did not come back fully readable.

Two returned nothing. Not a thin page, not a weak headline, nothing for a non-rendering client to quote. We wrote up the full metadata picture from that sample in our piece on what Tamil Nadu college homepages get wrong.

Want to know which of these applies to your site? Here is what a free AI visibility audit covers.

How do you check what a crawler sees on your page?

Three checks, in order of how much they tell you, and none of them needs a paid tool.

• 1

  Fetch the page with curl and no JavaScript engine, then read what comes back

• 2

  Open the rendered HTML in Search Console URL Inspection and compare it against that raw fetch

• 3

  Disable JavaScript in Chrome DevTools, reload, and look at the page as a non-rendering client would

"If your headline, your body copy and your enquiry path do not all survive a raw fetch, you are relying on rendering that only one of four platforms documents."

Absence from an AI answer has two possible causes, and this is only one of them. The diagnostic for separating them is in our note on why a site goes missing from AI answers.

Should you stop using JavaScript?

No. The question is not whether you use JavaScript, it is whether your content depends on it. A React site that server-renders its pages is readable by every path in our count. A React site that ships an empty shell and fills it on the client is readable by one.

"The web fetch tool currently does not support websites dynamically rendered with JavaScript."

That sentence is from Anthropic's web fetch tool documentation. It is the only unambiguous negative in our count, and it is worth more than the sixteen silences, because it tells you exactly what a reader on that platform gets when your page needs a browser. The fix is the same in every case: put the words in the HTML.