Weekly report · 2004-W50

The state of AI crawling

Measured on 30,472 reachable domains for the week ending 2004-12-12. Every figure is the result of an automated audit run the same way on every domain: read robots.txt, resolve the rules for 48 tracked agents, then send live requests as four of them.

← 2004-W49  ·  2004-W51 →

30,472
domains measured
7%
block an answer engine
20%
refuse AI agents at the edge
14%
publish an llms.txt

Block rate by crawler with change on the previous week

CrawlerOperatorContent used forBlock rateSitesWeek on week
GPTBot OpenAI training 7.1% 2,174 of 30,468 new
CCBot Common Crawl Foundation archive 6.8% 2,068 of 30,468 new
Bytespider ByteDance training 6.5% 1,988 of 30,468 new
ClaudeBot Anthropic training 5.8% 1,773 of 30,468 new
Amazonbot Amazon training 4.8% 1,467 of 30,468 new
Applebot-Extended Apple training 4.8% 1,453 of 30,468 new
Meta-ExternalAgent Meta training 4.8% 1,452 of 30,468 new
Google-Extended Google training 4.7% 1,440 of 30,468 new
anthropic-ai Anthropic training 4.4% 1,350 of 30,468 new
omgilibot Webz.io training 4.4% 1,345 of 30,468 new
Diffbot Diffbot training 4.2% 1,276 of 30,468 new
cohere-ai Cohere live retrieval 4.1% 1,264 of 30,468 new
omgili Webz.io training 4.0% 1,233 of 30,468 new
SemrushBot-OCOB Semrush training 3.7% 1,124 of 30,468 new
ChatGPT-User OpenAI live retrieval 3.4% 1,039 of 30,468 new
PerplexityBot Perplexity answer index 3.4% 1,039 of 30,468 new
FacebookBot Meta training 3.2% 966 of 30,468 new
YouBot You.com answer index 3.0% 901 of 30,468 new
ImagesiftBot ImageSift (Hive) training 2.8% 843 of 30,468 new
Timpibot Timpi training 2.7% 830 of 30,468 new
Scrapy Zyte (open-source framework) training 2.4% 727 of 30,468 new
AI2Bot Allen Institute for AI training 2.3% 705 of 30,468 new
OAI-SearchBot OpenAI answer index 2.1% 632 of 30,468 new
cohere-training-data-crawler Cohere training 2.0% 619 of 30,468 new
Meta-ExternalFetcher Meta live retrieval 1.7% 533 of 30,468 new
DuckAssistBot DuckDuckGo live retrieval 1.7% 527 of 30,468 new
PanguBot Huawei training 1.7% 519 of 30,468 new
Claude-User Anthropic live retrieval 1.6% 501 of 30,468 new
Claude-SearchBot Anthropic answer index 1.6% 480 of 30,468 new
Google-CloudVertexBot Google live retrieval 1.6% 478 of 30,468 new
Perplexity-User Perplexity live retrieval 1.6% 477 of 30,468 new
MistralAI-User Mistral AI live retrieval 1.5% 468 of 30,468 new
Kangaroo Bot Kangaroo LLM training 1.5% 454 of 30,468 new
GoogleOther Google training 1.4% 436 of 30,468 new
Meta-WebIndexer Meta answer index 1.2% 362 of 30,468 new
Applebot Apple answer index 1.2% 354 of 30,468 new
TikTokSpider ByteDance live retrieval 0.9% 261 of 30,468 new
Amzn-SearchBot Amazon answer index 0.8% 243 of 30,468 new
MistralAI-Training Mistral AI training 0.7% 206 of 30,468 new
Amzn-User Amazon live retrieval 0.6% 182 of 30,468 new
MistralAI-Index Mistral AI answer index 0.3% 93 of 30,468 new
facebookexternalhit Meta live retrieval 0.2% 51 of 30,468 new
Bingbot Microsoft answer index 0.1% 38 of 30,468 new
Googlebot-News Google answer index 0.1% 36 of 30,468 new
Meta-ExternalAds Meta training 0.1% 24 of 30,468 new
msnbot Microsoft answer index 0.1% 19 of 30,468 new
Googlebot Google answer index 0.0% 15 of 30,468 new
OAI-AdsBot OpenAI live retrieval 0.0% 12 of 30,468 new

Trend share of measured sites

% of sites20151050% of sites08-2209-2010-19block an answer engineblock a training crawlerpublish llms.txt

Where the blocking is share blocking an answer engine, by top-level domain

bar chart.uk (n=449)16.3%.de (n=842)16%.fr (n=335)14.9%.jp (n=421)8.8%.com (n=14927)6.8%.br (n=382)6%.org (n=1668)5.5%.net (n=1312)4.4%.in (n=311)3.2%.ru (n=1153)2.8%.edu (n=466)2.4%.io (n=654)2.1%

Only TLDs with at least 40 measured domains are shown, and each bar carries its sample size. A TLD is a weak proxy for jurisdiction. Full segmentation.

Changes recorded this week 0 total

No crawler policy changes were recorded in this week.

Full change log · RSS

Cite this CC BY 4.0

The figures on this page are frozen to the week shown and will not be edited after publication. Corrections are appended, not overwritten.

Crawl Census, "The state of AI crawling, 2004-W50", 2004-12-12. https://crawlcensus.com/report/2004-W50

Machine-readable: census totals, full corpus, crawler registry. Method: methodology.