All answers

The Prevailing
Answers for
Business model:
Business model, pick up to two

Do AI assistants' search crawlers (OpenAI, Anthropic, Perplexity) run JavaScript, and what must be in a page's raw HTML for them to read and cite it?

The PrevailingPublished Nov 17, 2026

Our answer

Mostly no, so put key facts in the raw HTML. ChatGPT, Claude and Perplexity read what your server sends and rarely run scripts. Render product names, prices, stock, reviews and key answers on the server, or prerender them.

Confidence
High
Strongest proof
3 controlled tests
5 official rules, 2 practitioners, 2 opinions or claims
Sources
10 sources
12 statements

New here? Take the 1-minute tour

Why this is the answer

3 reasons and our judgment

US assistants read only the raw HTML when they open a page for a user. In a test of 12, ChatGPT, Claude, Gemini, Perplexity, Meta AI and Copilot all reported a decoy, not the value a script wrote (Search Engine World, an SEO news site).

Their crawlers mostly skip scripts too. Over 80 days of beacon tests, OAI-SearchBot, PerplexityBot, ClaudeBot, ChatGPT-User and Claude-User never ran a page's JavaScript (JS SEO Lab, a bot rendering test lab). OpenAI's search crawler has since run scripts on some sites (Genezio, an analytics firm).

What goes missing is what sells. Airtable's plan prices and Ruggable's product prices appeared only after JavaScript ran, and Glossier's 7,150 reviews were not in the HTML, in a check of 40 well-known pages (Pritam Roy, a Fibr.ai cofounder).

Our judgment

We’d treat every one of these assistants as reading only your raw HTML, even though OpenAI’s search crawler has run scripts on some sites. The fetch that answers a user’s question did not run them in any test, and no operator documents rendering. We’d check prices, stock and reviews first, since those were the parts missing from raw HTML in Pritam Roy’s check and the lines geotoolbox says scripts add later.

Our judgment, built from Search Engine World, JS SEO Lab, Genezio, geotoolbox, OpenAI, Anthropic and Pritam Roy: our reading of these sources, not a finding from any one of them.

What the sources say

12 statements from 10 sources
  • AnthropicNov 17, 2026Supports

    Anthropic's API docs say Claude's web fetch tool does not support sites rendered dynamically with JavaScript. This covers the developer tool, not Claude-User or Claude-SearchBot.

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

    Read it on platform.claude.com

  • OpenAINov 17, 2026Supports

    OpenAI says OAI-SearchBot surfaces sites in ChatGPT search, while ChatGPT-User visits pages when users ask and robots.txt may not apply. Its page says nothing about JavaScript or rendering.

    “ChatGPT-User is not used for crawling the web in an automatic fashion.”

    Read it on developers.openai.com

  • OpenAIOct 28, 2026Supports

    OpenAI says sites opted out of OAI-SearchBot won't appear in ChatGPT search answers, so allow it in robots.txt. GPTBot is a separate setting for training, and changes take about 24 hours.

    “Sites that are opted out of OAI-SearchBot will not be shown in ChatGPT search answers, though can still appear as navigational links.”

    Read it on platform.openai.com

  • AnthropicOct 28, 2026Supports

    Anthropic says blocking Claude-SearchBot or Claude-User may reduce a site's visibility in Claude's search results; ClaudeBot is the separate training crawler.

    “Disabling Claude-SearchBot on your site prevents our system from indexing your content for search optimization, which may reduce your site's visibility and accuracy in user search results.”

    Read it on support.claude.com

  • PerplexityOct 28, 2026Supports

    Perplexity says to allow PerplexityBot in robots.txt and your firewall to appear in its search results; it is not used to train AI models, and its user-requested fetcher generally ignores robots.txt.

    “Since a user requested the fetch, this fetcher generally ignores robots.txt rules.”

    Read it on docs.perplexity.ai

  • Pritam RoyOct 22, 2026Supports

    Pritam Roy fetched 40 hand-picked pages with GPTBot's user agent: of 30 that served it, the median had 97% of rendered text in raw HTML, but web apps sent under 30% and some prices appeared only after JavaScript.

    “Look for price, stock, plan limits, reviews and dates. Those were the parts missing in my test.”

    Read it on pritamr.substack.com

  • GenezioOct 22, 2026Disagrees

    Genezio, an AI visibility vendor, says its own logs and a beacon test show OpenAI's search crawler runs JavaScript. It can't say how often, and still advises rendering important content on the server.

    “What changes is that a JS-only site is no longer invisible by default.”

    Read it on genezio.com

  • geotoolboxOct 4, 2026Caveat

    The vendor geotoolbox reviews others' logs: ChatGPT-User, Claude-User and PerplexityBot fetched no scripts, while OAI-SearchBot seems to render some pages after a delay.

    “The cheapest fix is the same either way: get the text into the first HTML response.”

    Read it on geotoolbox.ai

  • VercelAug 28, 2026Supports

    Vercel and ora.ai, both selling related services, found a fetch-only client could not get answers available only through JavaScript, and llms.txt or JSON-LD did not change what clients retrieved.

    “In the controlled probes, a fetch-only client could not retrieve an answer available only through JavaScript.”

    Read it on vercel.com

  • JS SEO LabAug 6, 2026Supports

    JS SEO Lab's 80-day beacon test on its own pages found ClaudeBot, PerplexityBot, OAI-SearchBot, ChatGPT-User and others never ran JavaScript, and GPTBot ran only inline scripts. Server-rendered HTML reached all.

    “In the server-rendered HTML: seen by everyone.”

    Read it on jsseo.dev

  • HybridRankingJun 14, 2026Supports

    HybridRanking, an AI visibility vendor, says to keep a page's main content (headline, body, product description, schema) in the raw HTML and check it with View Source. JavaScript is not forbidden.

    “View Source (Ctrl/Cmd-U) - the raw HTML the server returned, exactly what a non-rendering AI crawler sees.”

    Read it on hybridranking.com

  • Search Engine WorldJun 9, 2026Supports

    In a decoy test on one site, 7 of 12 AI assistants, including ChatGPT, Claude, Gemini and Perplexity, read only the raw HTML of a pasted page. It tests user-triggered fetches, not background crawlers.

    “Put your key facts, claims and numbers in the raw HTML, high up”

    Read it on searchengineworld.com

In the full answer

For readers who sign in
  • What to do this week3 steps
  • The numbers behind it3 figures from cited sources
  • Where each source standsA chart placing each source on the question
  • The strongest disagreementThe best case against the answer, from a cited source
  • How sure we areWhat its confidence rests on, and what would change the answer

Read next

More in AI search visibility