ElevenlabsCrawler has suddenly become one of the busiest bots in our logs. What is it, should we block it in robots.txt, and will blocking it stop ElevenLabs voice agents that customers point at our site from reading our pages? Can we tell whether the requests are really from ElevenLabs?

What should I check, step by step?

1. Find both tokens in your logs separately: "ElevenlabsCrawler" (training; full string Mozilla/5.0 (compatible; ElevenlabsCrawler/1.0; +https://elevenlabs.io/docs/bots)) and "ElevenlabsBot" (knowledge-base fetches; ElevenlabsBot/1.0). ElevenLabs says the version numbers may change, so match the token, not the whole string.

2. To opt out of training, add "User-agent: ElevenlabsCrawler" then "Disallow: /" (or only the sections you want excluded). Allow about 24 hours for ElevenLabs to pick it up.
3. If you want customers' ElevenLabs voice agents and receptionists to be able to learn from your public pages (support docs, pricing, opening hours), leave ElevenlabsBot allowed, and check that a "User-agent: *" Disallow isn't catching it: ElevenLabs says pages disallowed for all user agents are skipped too.
4. If you block ElevenlabsBot on purpose, expect customers to see failed imports; ElevenLabs tells them to check that the site allows its crawler user agent, and its receptionist docs suggest adding single pages or uploading the content as a file instead.
5. Check your CDN, WAF or bot-protection rules as well as robots.txt. ElevenLabs asks sites behind them to allow ElevenlabsBot, so a firewall rule can block it even when robots.txt allows it.
6. Treat the user-agent string as a claim, not proof. With no published IP list, you can't separate genuine ElevenLabs requests from a scraper using the same name. If volume is the problem, rate-limit by IP and path at your server or CDN rather than relying on robots.txt.
7. Watch for bursts. Independent observers have reported ElevenlabsCrawler sending thousands of requests to a site in a single day, from many hosting networks, then dropping to a trickle.
8. Keep a dated copy of each robots.txt version, so you can show when you opted out if your content's use for training is ever in question.

What is the difference between ElevenlabsCrawler and ElevenlabsBot?

ElevenLabs' bots page says it uses web crawlers to collect publicly available content for its AI models and to fetch pages for its products, and that site owners can control access with two robots.txt tokens, each independent of the other. ElevenlabsCrawler crawls the web to collect content that may be used to train ElevenLabs' generative AI models; disallowing it indicates your content should not be used for training.

• ElevenlabsBot fetches pages that ElevenLabs customers add to their agents' knowledge bases, when they add them and when they refresh them, and it honors robots.txt. The knowledge-base docs say URL imports and crawl jobs (whole-site crawls following links from a start URL, or importing from your sitemap) fetch with ElevenlabsBot/1.0, and that pages disallowed for ElevenlabsBot or for all user agents are skipped.
• So the two do different jobs for different people. ElevenlabsCrawler serves ElevenLabs' own models; ElevenlabsBot acts for an ElevenLabs customer who chose to point an agent at your pages, which may be one of your own customers, partners or you.

Why is ElevenlabsCrawler sending so many requests, and can you verify it's really ElevenLabs?

Several sites have recently reported ElevenlabsCrawler near the top of their crawler counts. One independent bot tracker reported 8,675 requests to two of its sites between October 19 and 26, 2026, more than any other crawler in the period (ClaudeBot had 4,937), with 8,347 of them on October 22 alone, nearly all answered with a 200, from about ten networks, mostly hosting providers. It noted that it couldn't tell whether those requests were ElevenLabs.

• Rightshold's own server logs show the same pattern: 842 requests with the ElevenlabsCrawler user agent between noon and 6 PM ET on November 10, 2026, more than any other crawler in that window. We can't verify that they came from ElevenLabs either.
• That gap is the practical problem. Google, Apple, Amazon and OpenAI publish IP lists or reverse-DNS patterns so you can check that a request claiming to be their crawler is genuine. ElevenLabs' bots page publishes neither, so a firewall can only match the name, and a block or allow rule on the name applies equally to anyone who copies it.

What happens if you block ElevenlabsCrawler or ElevenlabsBot in robots.txt?

Blocking ElevenlabsCrawler: by ElevenLabs' own description, it's the training crawler and it's independent of ElevenlabsBot, so the visible cost is close to none. It sends no visitors back. What you give up is having your content considered for training ElevenLabs' models.

• Blocking ElevenlabsBot, by name or by a blanket "User-agent: *" rule: ElevenLabs customers can't import or refresh your pages into their agents' knowledge bases. For a business whose customers, resellers or partners build voice agents or receptionists that answer questions about it, that can mean their agents work from stale or missing information.
• Neither line is a lock. robots.txt records a preference that a cooperating crawler follows; a scraper using either name doesn't have to. Content that must not be read belongs behind a login, and heavy traffic is a job for rate limits.