The AI bots, ranked by unwelcome

Every AI-associated user agent we check, and how many of the 649 readable sites block it. Snapshot: 2026-09-13

BotOperatorPurposeBlocked by
GPTBotOpenAItraining141 (21.7%)
ChatGPT-UserOpenAIfetch96 (14.8%)
OAI-SearchBotOpenAIsearch78 (12%)
ClaudeBotAnthropictraining146 (22.5%)
Claude-UserAnthropicfetch101 (15.6%)
Claude-SearchBotAnthropicsearch95 (14.6%)
Google-ExtendedGoogletraining126 (19.4%)
Applebot-ExtendedAppletraining128 (19.7%)
PerplexityBotPerplexitysearch123 (19%)
Perplexity-UserPerplexityfetch104 (16%)
CCBotCommon Crawltraining164 (25.3%)
BytespiderByteDancetraining163 (25.1%)
meta-externalagentMetatraining138 (21.3%)
AmazonbotAmazontraining121 (18.6%)
DuckAssistBotDuckDuckGosearch97 (14.9%)
MistralAI-UserMistralfetch97 (14.9%)
anthropic-aiAnthropictraining119 (18.3%)
Claude-WebAnthropicfetch119 (18.3%)
cohere-aiCoheretraining130 (20%)
cohere-training-data-crawlerCoheretraining100 (15.4%)
DiffbotDiffbottraining133 (20.5%)
omgiliWebz.iotraining130 (20%)
omgilibotWebz.iotraining114 (17.6%)
YouBotYou.comsearch114 (17.6%)
ImagesiftBotHivetraining116 (17.9%)
TimpibotTimpisearch113 (17.4%)
Meta-ExternalFetcherMetafetch96 (14.8%)
AI2BotAllen Institute for AItraining99 (15.3%)
DeepSeekBotDeepSeektraining81 (12.5%)
GrokBotxAItraining67 (10.3%)
PanguBotHuaweitraining96 (14.8%)
Gemini-Deep-ResearchGooglefetch73 (11.2%)

Purposes: training — collects data that may train models · search — builds an AI search index · fetch — retrieves a page live at a user's request. Sites are notably more willing to block training bots than fetch bots.