Block AI crawlers in WordPress without disappearing from AI search

Q: Will I disappear from ChatGPT, Claude or Perplexity answers?

Not with the recommended policy. It allows AI search and assistant bots that fetch pages to answer questions and cite sources, and blocks only training crawlers.

Velbrake shows which AI bots read your site and how much bandwidth they take. Block the crawlers that train on your content, keep the AI search bots that cite you, and catch scrapers that pretend to be Googlebot or GPTBot.

Download velbrake.zip (free, v1.0.0)

WordPress 6.2+ · PHP 7.4+ · GPL-2.0-or-later · No account, no tracking

Velbrake Pro

An add-on for the free plugin, unlocked with a licence key: protected paths, an IP allowlist, an /llms.txt file for AI search, a weekly email report, CSV export of bot traffic, and rule import/export to copy one site's setup to others. Free Velbrake keeps working without it.

Buy Personal, 1 site: $39/year Business, 5 sites: $79/year Agency, 25 sites: $149/year Lifetime, 25 sites: $349 once

Checkout, tax and receipts by Polar.sh. You get the plugin zip and your licence key right after payment.

What it does

Bandwidth report

Requests and data served per bot over 7, 30 or 90 days, plus what blocking saved. Enter your hosting price per GB to see it in dollars.

"Get cited, not copied"

One click blocks AI training crawlers (GPTBot, ClaudeBot, CCBot, Bytespider, Meta-ExternalAgent…) and keeps AI search bots (OAI-SearchBot, ChatGPT-User, PerplexityBot…).

Impostor detection

Checks bots against the official IP lists from OpenAI, Anthropic, Perplexity, Google and Microsoft, and reverse DNS for Apple and Amazon. Fakes get a 403.

Crawl-trap shield

Keeps bots off expensive URLs such as ?add-to-cart=, site search, filters, cart and checkout. Search engines are not affected.

Rate limits + robots.txt

Let a bot in, but no more than N requests a minute. Velbrake adds matching rules to robots.txt and never removes your lines.

Safe by default

Proxy-safe visitor IPs, no lock-outs when verification lists are stale, one regex check for human visitors, and it stores daily counts only, never IPs or URLs.

Screenshots

Velbrake report: requests, bandwidth served, requests stopped, impostors caught — The report: requests, bandwidth served, requests stopped and impostors caught.

Every bot seen with type, cost, identity check and a per-bot rule — Every bot seen, with its type, cost and identity check, and a per-bot rule.

One-click policies and per-group settings — One-click policies, then fine-tune each group of bots.

Protection settings: impostor blocking, crawl-trap shield, rate limit, robots.txt — Protection settings: impostor blocking, crawl-trap shield, rate limit and robots.txt.

Install in two minutes

Download velbrake.zip (don't unzip it).
In WordPress: Plugins → Add New → Upload Plugin, choose velbrake.zip, press Install Now, then Activate.
Open Settings → Velbrake. The recommended policy, Get cited, not copied, is already on.
Not sure yet? Choose Monitor only for a week and decide from real numbers.
Behind Cloudflare? Set "Visitor IP comes from" to Cloudflare. Behind another proxy, pick its header and add the proxy's IPs under "Trusted proxies".

Tested end-to-end in real WordPress (WordPress Playground) on PHP 8.3 and PHP 7.4: activation, settings, bot blocking, impostor checks, proxy handling and robots.txt.

FAQ

Will this hurt my Google rankings?

No. Search engines are allowed by default and the crawl-trap shield never applies to them. Blocking Google-Extended only opts you out of Gemini training; it does not affect Google Search.

Will I disappear from ChatGPT, Claude or Perplexity answers?

Not with the recommended policy. It allows the AI search and assistant bots that fetch pages to answer questions and cite sources, and blocks only the training crawlers.

Isn't robots.txt enough?

robots.txt is a request, not a lock. Velbrake writes it for you and also refuses blocked bots on the server, so the rules hold even when a bot ignores robots.txt.

I use a caching plugin. Does it still work?

Pages served from a disk or CDN cache never reach WordPress, so Velbrake can't see those requests. Most caches don't serve query-string URLs like ?add-to-cart= or ?s=, so the crawl-trap shield still covers the most expensive requests.

Can developers change a decision?

Yes, with the velbrake_decision filter; proxy lists via velbrake_trusted_proxies and velbrake_cloudflare_ranges.

What's the license?

GPL-2.0-or-later, like WordPress itself. Free to use on any number of sites. A Pro add-on is planned.