AI crawler study · AI companies
Which AI companies sites block AI crawlers?
6 of 44 AI companies sites block at least one AI crawler (13.6%). Checked October 2026, 45 sites.
Blocked most often
| Crawler | Kind | Sites blocking it |
|---|---|---|
| Google-Extended | AI-use control | 4 of 44 |
| Bytespider ByteDance | Training | 4 of 44 |
| OAI-SearchBot OpenAI | Search | 3 of 44 |
| ChatGPT-User OpenAI | User fetcher | 3 of 44 |
| PerplexityBot Perplexity | Search | 3 of 44 |
| CCBot Common Crawl | Training | 3 of 44 |
| Amazonbot Amazon | General | 3 of 44 |
| GPTBot OpenAI | Training | 2 of 44 |
| ClaudeBot Anthropic | Training | 2 of 44 |
| Claude-SearchBot Anthropic | Search | 2 of 44 |
| Claude-User Anthropic | User fetcher | 2 of 44 |
| Perplexity-User Perplexity | User fetcher | 2 of 44 |
| Applebot-Extended Apple | AI-use control | 2 of 44 |
| Meta-ExternalAgent Meta | Training | 2 of 44 |
Every site
| Site | Blocks | Which |
|---|---|---|
| meta.ai | 14 of 14 | all of them |
| poe.com | 12 of 14 | OAI-SearchBot, ChatGPT-User, ClaudeBot, Claude-SearchBot, Claude-User, PerplexityBot, Perplexity-User, Applebot-Extended, CCBot, Meta-ExternalAgent, Amazonbot, Bytespider |
| chatgpt.com | 4 of 14 | PerplexityBot, Google-Extended, CCBot, Bytespider |
| claude.ai | 4 of 14 | GPTBot, OAI-SearchBot, ChatGPT-User, Google-Extended |
| descript.com | 2 of 14 | Amazonbot, Bytespider |
| together.ai | 1 of 14 | Google-Extended |
| anthropic.com | 0 of 14 | none |
| cerebras.ai | 0 of 14 | none |
| character.ai | 0 of 14 | none |
| cohere.com | 0 of 14 | none |
| copy.ai | 0 of 14 | none |
| cursor.com | 0 of 14 | none |
| deepmind.google | 0 of 14 | none |
| deepseek.com | 0 of 14 | none |
| elevenlabs.io | 0 of 14 | none |
| glean.com | 0 of 14 | none |
| grammarly.com | 0 of 14 | none |
| grok.com | 0 of 14 | none |
| groq.com | 0 of 14 | none |
| harvey.ai | 0 of 14 | none |
| heygen.com | 0 of 14 | none |
| huggingface.co | 0 of 14 | none |
| ideogram.ai | 0 of 14 | none |
| jasper.ai | 0 of 14 | none |
| langchain.com | 0 of 14 | none |
| leonardo.ai | 0 of 14 | none |
| lovable.dev | 0 of 14 | none |
| mistral.ai | 0 of 14 | none |
| openai.com | 0 of 14 | none |
| openrouter.ai | 0 of 14 | none |
| otter.ai | 0 of 14 | none |
| perplexity.ai | 0 of 14 | none |
| pinecone.io | 0 of 14 | none |
| replicate.com | 0 of 14 | none |
| replit.com | 0 of 14 | none |
| runwayml.com | 0 of 14 | none |
| scale.com | 0 of 14 | none |
| stability.ai | 0 of 14 | none |
| suno.com | 0 of 14 | none |
| synthesia.io | 0 of 14 | none |
| wandb.ai | 0 of 14 | none |
| windsurf.com | 0 of 14 | none |
| x.ai | 0 of 14 | none |
| you.com | 0 of 14 | none |
Left out, because their robots.txt could not be read (bot protection, errors or no answer): midjourney.com.
From each site's robots.txt rules for the homepage, read as the robots.txt standard (RFC 9309) says. robots.txt is a request, not a lock: this shows what sites ask, not what AI companies do. Sites whose robots.txt we could not read are left out. The list of sites was picked by Kavel and is not a sample of the web. Part of Kavel's AI crawler study, built with AI.
What does your site allow? Check any site live, crawler by crawler.