# BotLedger > Which AI bots really visit your site? Check your access log against the official IP ranges of OpenAI, Anthropic, Perplexity, Google and more — and see how much of your “AI traffic” is fake. ## Tools - [API examples](https://botledger.website/api-docs/examples): tested middleware for Python, Express and PHP that blocks fake AI crawlers via the API. - [AI bot verification API](https://botledger.website/api-docs): REST API — is this IP + user agent really GPTBot/ClaudeBot/…? and Web Bot Auth signature verification. Free 10k calls/month, Pro $19/month for 1M. - [Free AI-bot log checker](https://botledger.website/#check): checks every AI-bot request in an nginx/Apache access log against the operators' official IP ranges, in the browser. - [Free WordPress plugin](https://botledger.website/wordpress-plugin): verified AI-bot analytics inside WordPress (real vs fake per bot, pages AI reads, AI referrals, optional fake-bot blocking). - [Web Bot Auth explained](https://botledger.website/guides/web-bot-auth): how AI agents sign requests and which ones publish key directories (Google-Agent agent.bot.goog, ChatGPT chatgpt.com, Browserbase, You.com), refreshed nightly. - [Web Bot Auth validator](https://botledger.website/tools/web-bot-auth-validator): check an AI agent's signed request (Signature, Signature-Input, Signature-Agent) against the IETF draft, incl. signature base and Ed25519/RSA-PSS verification; in-browser. - [Machine-readable feed](https://botledger.website/data/v1/ranges.json): JSON with every bot (UA regex, robots token) and all official IP prefixes, refreshed daily. - [Block fake AI bots in nginx](https://botledger.website/guides/block-fake-ai-bots-nginx): drop-in config ([raw](https://botledger.website/data/v1/nginx-fake-ai-bots.conf)) that returns 403 when a request claims an AI bot user agent from an IP outside the operator's official list; real bots pass. - [Block fake AI bots with .htaccess / Apache](https://botledger.website/guides/block-fake-ai-bots-apache): same check as mod_rewrite rules ([raw](https://botledger.website/data/v1/htaccess-fake-ai-bots.txt)). - [Block fake AI bots on Cloudflare](https://botledger.website/guides/block-fake-ai-bots-cloudflare): one free-plan WAF custom rule (UA claims an AI bot and not cf.client.bot → block). - [Block fake AI bots on WordPress](https://botledger.website/guides/block-fake-ai-bots-wordpress): free BotLedger plugin (IP check vs official lists, 403 for fakes) or server/Cloudflare rules when page caching bypasses PHP; robots.txt for the real bots. - [AI robots.txt generator](https://botledger.website/tools/robots-txt-generator): block AI training bots while staying in AI search, or block/allow all; tokens from operator docs. ## MCP server - URL: https://botledger.website/api/aibots/mcp (Streamable HTTP, no sign-in) - Tools: list_ai_bots, verify_bot_ip, robots_txt_rules, get_api_key - Descriptor: https://botledger.website/.well-known/mcp.json - Connect steps for Claude, ChatGPT and Claude Code: https://botledger.website/for-ai-agents#mcp ## Research - [At least 43% of AI-bot traffic is fake](https://botledger.website/research/fake-ai-bot-traffic): 15-day log study, methodology and limits. - [Which AI crawlers do the top 1,000 websites block?](https://botledger.website/research/ai-crawler-blocking): monthly robots.txt study (Tranco top 1k): ~24% block at least one AI training crawler, ~11% block ChatGPT search, 3.6% block everything; dataset JSON. ## AI crawler directory - [GPTBot](https://botledger.website/bots/gptbot) — OpenAI, Model training crawler. 18 official IP prefixes. - [OAI-SearchBot](https://botledger.website/bots/oai-searchbot) — OpenAI, AI search indexer. 39 official IP prefixes. - [ChatGPT-User](https://botledger.website/bots/chatgpt-user) — OpenAI, User-triggered fetcher. 237 official IP prefixes. - [OAI-AdsBot](https://botledger.website/bots/oai-adsbot) — OpenAI, Ads validation. 2 official IP prefixes. - [ClaudeBot](https://botledger.website/bots/claudebot) — Anthropic, Model training crawler. 14 official IP prefixes. - [Claude-SearchBot](https://botledger.website/bots/claude-searchbot) — Anthropic, AI search indexer. 14 official IP prefixes. - [Claude-User](https://botledger.website/bots/claude-user) — Anthropic, User-triggered fetcher. 14 official IP prefixes. - [PerplexityBot](https://botledger.website/bots/perplexitybot) — Perplexity, AI search indexer. 13 official IP prefixes. - [Perplexity-User](https://botledger.website/bots/perplexity-user) — Perplexity, User-triggered fetcher. 10 official IP prefixes. - [Google-Extended](https://botledger.website/bots/google-extended) — Google, robots.txt token only. No HTTP user agent (robots.txt token only). - [Googlebot](https://botledger.website/bots/googlebot) — Google, AI search indexer. 2181 official IP prefixes. - [Bingbot](https://botledger.website/bots/bingbot) — Microsoft, AI search indexer. 29 official IP prefixes. - [Applebot](https://botledger.website/bots/applebot) — Apple, AI search indexer. 24 official IP prefixes. - [CCBot](https://botledger.website/bots/ccbot) — Common Crawl, AI / data crawler. 5 official IP prefixes. - [DuckAssistBot](https://botledger.website/bots/duckassistbot) — DuckDuckGo, User-triggered fetcher. 486 official IP prefixes. - [MistralAI-User](https://botledger.website/bots/mistralai-user) — Mistral AI, User-triggered fetcher. 4 official IP prefixes. - [Bytespider](https://botledger.website/bots/bytespider) — ByteDance, AI / data crawler. No official IP list. - [meta-externalagent](https://botledger.website/bots/meta-externalagent) — Meta, Model training crawler. No official IP list. - [meta-externalfetcher](https://botledger.website/bots/meta-externalfetcher) — Meta, User-triggered fetcher. No official IP list. - [Amazonbot](https://botledger.website/bots/amazonbot) — Amazon, Model training crawler. No official IP list. - [Amzn-SearchBot](https://botledger.website/bots/amzn-searchbot) — Amazon, AI search indexer. No official IP list. - [Google-CloudVertexBot](https://botledger.website/bots/google-cloudvertexbot) — Google, AI / data crawler. 2181 official IP prefixes. ## Key facts - User agents can be spoofed; verify AI crawlers by source IP against the operator's published list. - Anthropic publishes its bot IPs at https://claude.com/crawling/bots.json. - Google-Extended is a robots.txt token with no HTTP user agent; requests claiming it are fake. ## Markdown versions - Markdown index of every page: https://botledger.website/sitemap.md - Every public page has a plain Markdown version: add .md to its path (home: https://botledger.website/index.md).