AI crawler access on drupal sites
207 of the 867 sites measured in this segment refuse a request from an identified AI user agent at the edge, 24% of the segment, against 4% whose robots.txt disallows an answer engine at the site root. The block that costs these sites citations is not the one written in the file. That answer-engine rate is 2.8 points below the corpus rate of 6%, which is itself measured across 30,353 domains.
Mean AI access score across those 867 sites is 74.9 of 100, 50 of them publish an llms.txt (6%), and 275 publish structured data an answer engine can parse (32%).
Block rate by crawler, inside this segment robots.txt at the site root
| Crawler | Operator | Content used for | Block rate | Blocking | Measured |
|---|---|---|---|---|---|
| Bytespider | ByteDance | training | 7.3% | 63 | 867 |
| GPTBot | OpenAI | training | 6.9% | 60 | 867 |
| CCBot | Common Crawl Foundation | archive | 6.5% | 56 | 867 |
| ClaudeBot | Anthropic | training | 6.1% | 53 | 867 |
| Meta-ExternalAgent | Meta | training | 5.8% | 50 | 867 |
| Applebot-Extended | Apple | training | 5.8% | 50 | 867 |
| Google-Extended | training | 5.7% | 49 | 867 | |
| Amazonbot | Amazon | training | 5.4% | 47 | 867 |
| SemrushBot-OCOB | Semrush | training | 3.5% | 30 | 867 |
| anthropic-ai | Anthropic | training | 2.4% | 21 | 867 |
| ChatGPT-User | OpenAI | live retrieval | 2.3% | 20 | 867 |
| cohere-ai | Cohere | live retrieval | 2.2% | 19 | 867 |
| PerplexityBot | Perplexity | answer index | 2.1% | 18 | 867 |
| omgilibot | Webz.io | training | 2.1% | 18 | 867 |
| omgili | Webz.io | training | 2.0% | 17 | 867 |
| Diffbot | Diffbot | training | 2.0% | 17 | 867 |
| FacebookBot | Meta | training | 1.8% | 16 | 867 |
| OAI-SearchBot | OpenAI | answer index | 1.6% | 14 | 867 |
| YouBot | You.com | answer index | 1.5% | 13 | 867 |
| Scrapy | Zyte (open-source framework) | training | 1.5% | 13 | 867 |
| ImagesiftBot | ImageSift (Hive) | training | 1.4% | 12 | 867 |
| Timpibot | Timpi | training | 1.3% | 11 | 867 |
| Applebot | Apple | answer index | 1.2% | 10 | 867 |
| AI2Bot | Allen Institute for AI | training | 1.2% | 10 | 867 |
| cohere-training-data-crawler | Cohere | training | 1.0% | 9 | 867 |
| Meta-WebIndexer | Meta | answer index | 0.9% | 8 | 867 |
| PanguBot | Huawei | training | 0.8% | 7 | 867 |
| Meta-ExternalFetcher | Meta | live retrieval | 0.8% | 7 | 867 |
| Kangaroo Bot | Kangaroo LLM | training | 0.8% | 7 | 867 |
| TikTokSpider | ByteDance | live retrieval | 0.7% | 6 | 867 |
| GoogleOther | training | 0.7% | 6 | 867 | |
| facebookexternalhit | Meta | live retrieval | 0.7% | 6 | 867 |
| DuckAssistBot | DuckDuckGo | live retrieval | 0.7% | 6 | 867 |
| Claude-User | Anthropic | live retrieval | 0.7% | 6 | 867 |
| Claude-SearchBot | Anthropic | answer index | 0.7% | 6 | 867 |
| Google-CloudVertexBot | live retrieval | 0.6% | 5 | 867 | |
| Perplexity-User | Perplexity | live retrieval | 0.5% | 4 | 867 |
| MistralAI-User | Mistral AI | live retrieval | 0.5% | 4 | 867 |
| Amzn-SearchBot | Amazon | answer index | 0.5% | 4 | 867 |
| Bingbot | Microsoft | answer index | 0.3% | 3 | 867 |
| Amzn-User | Amazon | live retrieval | 0.2% | 2 | 867 |
| OAI-AdsBot | OpenAI | live retrieval | 0.1% | 1 | 867 |
| msnbot | Microsoft | answer index | 0.0% | 0 | 867 |
| MistralAI-Training | Mistral AI | training | 0.0% | 0 | 867 |
| MistralAI-Index | Mistral AI | answer index | 0.0% | 0 | 867 |
| Meta-ExternalAds | Meta | training | 0.0% | 0 | 867 |
| Googlebot-News | answer index | 0.0% | 0 | 867 | |
| Googlebot | answer index | 0.0% | 0 | 867 |
Block rate is the share of this segment's domains holding a policy record for that agent whose robots.txt disallows it at the site root, shown with its numerator and denominator. Denominators differ between agents because a record only exists once a domain has been measured for it. A domain serving no robots.txt counts as allowing everything, which is what the standard specifies.
Most accessible
| Domain | Score | Engines blocked | llms.txt | |
|---|---|---|---|---|
| 1 | sahistory.org.za | 99 | 0 | yes |
| 2 | wri.org | 98 | 0 | yes |
| 3 | posit.co | 97 | 0 | yes |
| 4 | skynettechnologies.com | 97 | 0 | yes |
| 5 | gameinformer.com | 96 | 0 | yes |
| 6 | belvedere.at | 96 | 0 | yes |
| 7 | mtholyoke.edu | 95 | 0 | no |
| 8 | starwoodhotels.com | 94 | 0 | yes |
| 9 | mindbodyonline.com | 94 | 0 | no |
| 10 | acquia.com | 94 | 0 | yes |
| 11 | monsido.com | 94 | 0 | yes |
| 12 | purple.com | 94 | 0 | yes |
| 13 | nea.org | 94 | 0 | no |
| 14 | flexera.com | 93 | 0 | no |
| 15 | tpg.com.au | 93 | 0 | no |
Highest AI access scores among the 867 sites measured in this segment.
Least accessible
| Domain | Score | Engines blocked | llms.txt | |
|---|---|---|---|---|
| 1 | hhmi.org | 45 | 0 | no |
| 2 | flgov.com | 47 | 0 | no |
| 3 | icj-cij.org | 47 | 0 | no |
| 4 | penguinrandomhouse.ca | 48 | 0 | no |
| 5 | myplate.gov | 49 | 0 | no |
| 6 | taxpolicycenter.org | 50 | 0 | no |
| 7 | nic.aero | 50 | 0 | no |
| 8 | fraserinstitute.org | 51 | 6 | no |
| 9 | colorado.gov | 51 | 0 | no |
| 10 | anses.fr | 53 | 0 | no |
| 11 | uscis.gov | 53 | 0 | no |
| 12 | safekids.org | 53 | 0 | no |
| 13 | nationalgrid.com | 54 | 0 | no |
| 14 | gites-de-france.com | 54 | 0 | no |
| 15 | usdoj.gov | 54 | 0 | no |
Lowest AI access scores among the same 867 sites.
Compare with the rest of the grouping publishing platform segments
The largest 10 other segments in this grouping, ordered by size. The share beside each is that segment's own answer-engine block rate over its own count. Full table on the segments index.
What this segment does not tell you read this before citing it
Platform detection relies on a generator meta tag, so coverage is partial and skewed. A site that strips the tag, or a system that never writes one, is invisible here. These rows therefore over-represent platforms that advertise themselves and say nothing at all about the ones that do not.
Measure your own domain
The audit that produced every number on this page runs on any domain in about six seconds: robots.txt resolved for 48 agents, then live requests sent as four of them to see whether the edge agrees with the file.