Every AI search engine gets its raw material from a different place. ChatGPT search uses OpenAI's crawler plus third-party search providers. Google AI Overviews and AI Mode use Google's search index. Naver AI Briefing leans on Naver's own blogs, cafes and Q&A. Perplexity and Claude build indexes with their own search crawlers. That's why a brand that looks strong in one engine can be missing from another.
At Anymorph we ask the same questions to several AI engines on the same day and compare the answers and cited sources. This piece maps how each engine sources its answers, based on what each company has published, and what the differences mean for brands. It's current as of October 2026.
Key takeaways
- ChatGPT search uses pages collected by OAI-SearchBot and results from third-party search providers. It has 1.2 billion weekly users (September 2026).
- Google AI Overviews and AI Mode pick sources from Google's index and run fan-out searches on sub-questions.
- The Gemini app grounds answers in Google Search but skips sites that block Google-Extended.
- Perplexity and Claude build their own indexes with search crawlers (PerplexityBot, Claude-SearchBot) and fetch pages live when users ask.
- Microsoft Copilot uses Bing's index, and Bing Webmaster Tools shows your citations.
- Naver AI Briefing takes 70% of its citations from Naver Blog, Cafe and Knowledge iN, which block the search bots of ChatGPT, Perplexity and Claude.
Engine-by-engine comparison
| Engine | Where answers come from | Crawlers to allow | Report for site owners |
|---|---|---|---|
| ChatGPT search | Pages from OAI-SearchBot, third-party search providers, partner content | OAI-SearchBot, Bingbot | None; track utm_source=chatgpt.com in analytics |
| Google AI Overviews, AI Mode | Google's index (RAG, query fan-out) | Googlebot; site included in Search Console's Search generative AI control | Search Console Generative AI performance report |
| Gemini app | Grounding in Google Search | Googlebot, Google-Extended | None |
| Perplexity | Its own index plus live fetches | PerplexityBot | None |
| Claude | Web search tool and its own search index | Claude-SearchBot, Claude-User | None |
| Microsoft Copilot | Bing's index | Bingbot | AI Performance in Bing Webmaster Tools |
| Naver AI Briefing | Naver search, Naver Blog, Cafe, Knowledge iN | Yeti; Naver Search Advisor registration | None |
A closer look
ChatGPT search
ChatGPT rewrites a question into one or more search queries, sends them to third-party search providers, and may follow up with more specific queries after reading the first results. OpenAI's example turns a question about drugs targeting CCR8 into the query "CCR8 immunotherapy drug development 2025". To be a candidate source, don't block OAI-SearchBot; blocked sites can still show up, but only as navigational links. ChatGPT adds utm_source=chatgpt.com to outbound links, so the traffic is easy to separate in Google Analytics.
Google AI Overviews and AI Mode
Google's core ranking systems retrieve relevant pages from the index; the AI reviews them, writes the answer and links to its sources. Because of query fan-out, a page that doesn't rank for the original query can still be cited if it answers a sub-question well. The details are in how to appear in Google AI Overviews and AI Mode.
The Gemini app
The Gemini app grounds its answers in Google Search. But if a site blocks Google-Extended in robots.txt, its pages are used neither for Gemini training nor for grounding in the Gemini app. A page can rank well on Google and still be absent from Gemini's answers for that reason. Alphabet reported 950 million monthly Gemini app users in its July 2026 earnings.
Perplexity
Perplexity finds candidates in the index built by PerplexityBot and opens pages live with Perplexity-User when needed. Every answer carries numbered citations. Perplexity says PerplexityBot isn't used for training, and that Perplexity-User, acting on a user's request, generally ignores robots.txt.
Claude
Claude added web search in March 2025 and opened it to all plans that May. Anthropic runs Claude-SearchBot to index pages for search quality and Claude-User to fetch pages when users ask. Blocking Claude-SearchBot, Anthropic says, may reduce your visibility in Claude's search results.
Microsoft Copilot
Copilot and Bing's AI summaries ground their answers in Bing's index. In February 2026 Microsoft opened an AI Performance report in Bing Webmaster Tools, showing which of your pages were cited across Copilot, Bing's AI summaries and some partner integrations, and the grounding queries behind them. It's a rare case of an AI engine handing citation data to site owners.
Naver AI Briefing
AI Briefing appeared on 27% of all Naver searches in July 2026, and since January 2026 it has taken 70% of its citations from Naver Blog, Cafe and Knowledge iN. Outside websites share the rest with news and official sources. Naver also runs a separate conversational search, AI Tab. See how Naver AI Briefing picks its sources.
Why the same brand looks different in each engine
Different candidate pools. Each engine uses a different index, and every site and platform opens different crawlers. When we checked on October 4, 2026, Naver Blog allowed Googlebot but blocked OAI-SearchBot, PerplexityBot, Claude-SearchBot and Google-Extended. A carefully written Naver Blog post can reach Naver AI Briefing and Google Search and still be unusable as a source for ChatGPT or the Gemini app. Our AI crawler guide has the full table for Korean platforms.
Rewritten queries. Google fans questions out into sub-queries, and ChatGPT rewrites them into search queries. Perplexity's Q2D-Web benchmark, published in September 2026, was built from 70,000 agent-reformulated queries in ten languages based on real user questions. The queries the AI writes, not the sentence the person typed, decide which pages become candidates.
Models and timing. Google says AI Overviews and AI Mode may use different models and techniques. Ask the same engine the same question on different days and the answer and citations shift. One check is never enough.
How to measure, engine by engine
- Fix your question list. Choose 30 to 100 questions customers would ask AI, and use the same list for every engine.
- Ask on the same day, under the same conditions. To compare engines, keep questions, date, language and location constant.
- Record three things. Whether your brand is mentioned, in what position, and which sources are cited. Do the same for competitors.
- Use the official reports too. Search Console's Generative AI performance report for Google, AI Performance in Bing Webmaster Tools for Copilot, and referral traffic from chatgpt.com, perplexity.ai, gemini.google.com and copilot.microsoft.com in your analytics.
- Repeat weekly. Trends on the same questions are more reliable than any single answer.
Anymorph Monitor asks the same questions every day to ChatGPT, Gemini, Google AI Overviews and AI Mode, Perplexity, and Naver AI Briefing and AI Tab, and shows mention rate, share of voice and cited sources next to the original answers.
FAQ
If I can only focus on one engine, which first?
Start where your customers actually ask. For consumers in Korea, that usually means Naver AI Briefing, ChatGPT and Google AI Overviews together. Their sourcing differs enough that winning one doesn't carry the others.
If I rank well on Google, will I show up in ChatGPT?
Not necessarily. ChatGPT search draws on OAI-SearchBot and third-party providers, so Google rankings and ChatGPT citations can diverge. Start by confirming OAI-SearchBot and Bingbot can reach your site.
AI answers change every time. Is measuring even meaningful?
That's exactly why you measure repeatedly. A single answer has noise in it. The same questions asked under the same conditions over several weeks show clear trends in mention rates and sources.
Sources
- OpenAI, Searching the web with ChatGPT
- OpenAI, Publishers and Developers FAQ
- OpenAI, Overview of OpenAI Crawlers
- The Verge, ChatGPT now has 1.2 billion weekly users, OpenAI says (September 2026)
- Google Search Central, Optimizing your website for generative AI features on Google Search
- Google, List of Google's common crawlers
- TechRepublic, Google Says Gemini Reaches 950 Million Monthly Users (July 2026)
- Perplexity, Perplexity Crawlers
- Perplexity, Q2D-Web: Evaluating First-Stage Retrievers at Scale (September 2026)
- Anthropic, Claude can now search the web (March 2025)
- Anthropic, Does Anthropic crawl data from the web?
- Microsoft Bing, Introducing AI Performance in Bing Webmaster Tools Public Preview (February 2026)
- NAVER, CDO Kim Kwang-hyun Reveals Data & Content Strategy for the AI Era (May 2026)