user_agent stringlengths 5 17 | operator stringclasses 7
values | category stringclasses 6
values | purpose stringlengths 35 194 | respects_robots_txt stringclasses 4
values | blocking_reduces_ai_search_visibility stringclasses 2
values | source_url stringclasses 7
values | last_verified stringdate 2026-10-28 00:00:00 2026-10-28 00:00:00 |
|---|---|---|---|---|---|---|---|
GPTBot | OpenAI | training | Crawls content that may be used to train OpenAI's generative AI foundation models | yes | no | https://developers.openai.com/docs/bots | 2026-10-28 |
OAI-SearchBot | OpenAI | search_index | Surfaces websites in ChatGPT search features | yes | yes | https://developers.openai.com/docs/bots | 2026-10-28 |
ChatGPT-User | OpenAI | user_fetch | Visits a page for user-initiated actions in ChatGPT and Custom GPTs; not used for automatic crawling | may_not_apply | yes | https://developers.openai.com/docs/bots | 2026-10-28 |
OAI-AdsBot | OpenAI | ads_check | Checks the safety of landing pages submitted as ads in ChatGPT; only visits submitted ad pages | not_stated | no | https://developers.openai.com/docs/bots | 2026-10-28 |
ClaudeBot | Anthropic | training | Collects public web content that could contribute to training Anthropic's generative AI models | yes | no | https://support.claude.com/en/articles/8896518 | 2026-10-28 |
Claude-SearchBot | Anthropic | search_index | Navigates the web to improve the quality and relevance of Claude's search results | yes | yes | https://support.claude.com/en/articles/8896518 | 2026-10-28 |
Claude-User | Anthropic | user_fetch | Accesses websites when a Claude user asks a question | yes | yes | https://support.claude.com/en/articles/8896518 | 2026-10-28 |
PerplexityBot | Perplexity | search_index | Surfaces and links websites in Perplexity search results; not used to crawl content for AI foundation models | yes | yes | https://docs.perplexity.ai/docs/resources/perplexity-crawlers | 2026-10-28 |
Perplexity-User | Perplexity | user_fetch | Visits a page to help answer a user's question and link to it; not used for crawling or training data | generally_ignored | yes | https://docs.perplexity.ai/docs/resources/perplexity-crawlers | 2026-10-28 |
Google-Extended | Google | robots_token | robots.txt token (no separate HTTP user agent) controlling whether Google-crawled content may be used to train Gemini models and for grounding; does not affect Google Search inclusion or ranking | yes | no | https://developers.google.com/crawling/docs/crawlers-fetchers/google-common-crawlers | 2026-10-28 |
Applebot | Apple | search_index | Apple's web crawler; powers search features in Spotlight, Siri and Safari and may also be used to train Apple's foundation models | yes | yes | https://support.apple.com/en-us/HT6619 | 2026-10-28 |
Applebot-Extended | Apple | robots_token | Secondary user agent that does not crawl pages; lets publishers opt out of Applebot-crawled content being used to train Apple's foundation models | yes | no | https://support.apple.com/en-us/HT6619 | 2026-10-28 |
CCBot | Common Crawl | open_dataset | Builds the free Common Crawl web archive used by researchers and companies, including for AI training | yes | no | https://commoncrawl.org/faq | 2026-10-28 |
Bingbot | Microsoft | search_index | Microsoft's crawler for Bing search | yes | yes | https://blogs.bing.com/webmaster/September-2010/Bingbot-is-coming-to-town | 2026-10-28 |
AI Crawler User Agents
A small reference dataset of the web crawlers and robots.txt tokens used by AI companies. For each one it records what the agent does, whether it follows robots.txt, and whether blocking it is likely to reduce a site's visibility in AI search answers. Every row links to the operator's own documentation.
Compiled and maintained by Smart Local, a local digital marketing and AI search agency in Glasshouse Mountains, Queensland. It is the reference list behind the open-source ai-search-readiness checker.
Files
| File | Format |
|---|---|
ai_crawlers.csv |
CSV, one row per user agent or token |
ai_crawlers.jsonl |
The same rows as JSON Lines |
Fields
| Field | Meaning |
|---|---|
user_agent |
The name used in robots.txt User-agent: lines |
operator |
Company or organisation that runs it |
category |
training (collects content for model training), search_index (indexes pages for a search or answer product), user_fetch (fetches a page because a user asked), robots_token (a robots.txt control with no crawler of its own), open_dataset (public web archive), ads_check |
purpose |
Short paraphrase of the operator's own description |
respects_robots_txt |
yes, may_not_apply / generally_ignored (operator says user-initiated fetches may skip robots.txt), or not_stated |
blocking_reduces_ai_search_visibility |
yes if the operator's documentation indicates that blocking it removes or reduces the site in that operator's search or answer results; no for training-only agents and tokens |
source_url |
Operator documentation the row was checked against |
last_verified |
Date the row was checked against the source |
Methodology
- Started from the user agents checked by ai-search-readiness, plus any agent named in the same operator documentation.
- Read each operator's official crawler documentation (linked in
source_url). Third-party lists were not used as sources. - Paraphrased each stated purpose and robots.txt behaviour. Where the operator does not say, the value is
not_statedrather than a guess. - Classified
blocking_reduces_ai_search_visibilityfrom the operator's wording. Training crawlers and opt-out tokens areno. Search indexers and user fetchers areyes. - All rows were last verified on 28 October 2026.
Limitations
- Operators add, rename and change agents. Check
source_urlbefore relying on a row. - "Respects robots.txt" is the operator's own statement. It is not an independent test of crawler behaviour.
- Coverage is limited to the agents most relevant to AI search visibility for websites. It is not a complete list of all bots.
- A crawler visiting a page is not proof that the page was cited in an AI answer. See the research note below.
Research note
How to tell AI crawler visits, AI citations and referral visits apart in your own server logs: https://smart-local.com.au/resources/ai-crawl-vs-citation/
Licence and citation
CC BY 4.0. Please credit:
Smart Local (2026). AI Crawler User Agents [Data set]. Hugging Face. https://huggingface.co/datasets/smart-local/ai-crawler-user-agents
Smart Local · 12 Clearview Drive, Glasshouse Mountains QLD 4518 · 1300 305 488 · https://smart-local.com.au/
- Downloads last month
- 12