AI Overviews and ChatGPT search look similar in a screenshot and behave nothing alike underneath. AI Overviews are generated inside Google Search, from pages Googlebot has already crawled and indexed. ChatGPT search pulls from OpenAI's own crawlers plus third-party partners. That single difference decides which work matters.
If you are treating "get cited by AI" as one project, you are probably spending effort on the wrong crawler for half your goal.
The plumbing, in plain terms
AI Overviews are part of Google Search. Google's documentation treats them as a Search feature, which means they are governed by normal Googlebot crawling and indexing. There is no separate opt-in, no separate file to allow, and no separate index. If a page is not indexed and eligible to rank, it is not a candidate for an AI Overview.
ChatGPT search is a different pipeline. OpenAI documents a set of user agents with distinct jobs:
- OAI-SearchBot surfaces websites in ChatGPT search. Sites that opt out of OAI-SearchBot are not shown in ChatGPT search answers.
- GPTBot collects content to improve foundation models. Blocking GPTBot does not affect ChatGPT search.
- ChatGPT-User handles user-initiated actions in ChatGPT and custom GPTs. Because a user triggered it, robots.txt rules may not apply, and it is not used to decide whether content appears in search.
So the ChatGPT side has a search crawler you can allow or block independently of the training crawler. The Google side does not offer that split for AI Overviews.
Google-Extended is not the AI Overviews switch
This is the most common mix-up I see. Google-Extended controls whether content is used to train future Gemini models and for grounding in Gemini apps and the Vertex AI Gemini API. It does not affect inclusion in Google Search, and it is not a ranking signal. You can read the details on Google's crawler documentation.
Blocking Google-Extended does nothing to your AI Overviews eligibility. If you blocked it and then wondered why you still appear in AI Overviews, that is expected behavior, not a bug.
AI Overviews are driven by Googlebot. That is the whole lever.
What each game rewards
| AI Overviews | ChatGPT search | |
|---|---|---|
| Crawler that matters | Googlebot | OAI-SearchBot (for search inclusion) |
| Underlying index | Google's index | OpenAI's own retrieval plus partners |
| Opt-out that removes you | None specific to AI Overviews | Opting out of OAI-SearchBot |
| Training control | Google-Extended (separate, no Search effect) | GPTBot (separate, no Search effect) |
| Change propagation | Normal recrawl timing | Robots.txt changes can take about 24 hours |
Two things fall out of that table. First, on the Google side you are still playing a ranking game, because AI Overviews draw from pages that rank. Second, on the ChatGPT side you are playing a retrieval-and-quoting game, because the system is assembling an answer from sources it can fetch and cite.
Those are related skills, but they are not the same checklist.
The levers that actually differ
For AI Overviews
Because AI Overviews sit on top of the Google index, the classic fundamentals still apply: crawlability, indexation, and ranking for the query. A page buried on page four is not a realistic source for an overview on that query.
What tends to help, in my experience:
- Make the page's core answer extractable. A clear, self-contained passage that answers the question beats a page that only answers it implicitly across eight sections.
- Keep the page eligible. No accidental noindex, no blocked Googlebot, no content hidden behind interaction that a crawler will not trigger.
- Cover the question, not just the keyword. Overviews are assembled from pages that address the query as asked.
What does not help: fiddling with Google-Extended. It is not connected.
For ChatGPT search
Here the first question is access. If you have opted out of OAI-SearchBot, you are not in the running, full stop. That is documented behavior. If you have blocked GPTBot but left OAI-SearchBot alone, you are still eligible for search while keeping content out of foundation-model training. That distinction is worth checking in your robots.txt, because plenty of sites blocked everything in one line and never revisited it. OpenAI's user agent list is on its bots documentation.
Beyond access, the work is about being quotable. ChatGPT search is assembling an answer, so passages that state a fact, a definition, or a recommendation cleanly are easier to lift than pages that make the reader assemble the point themselves.
One caution: ChatGPT-User is not a search-inclusion signal, and because it fires on user request, robots.txt may not govern it. Do not read ChatGPT-User hits in your logs as evidence that you are being cited.
A note on robots.txt mechanics
Under RFC 9309, a crawler follows the most specific group that names it and ignores the * group in that case. Within a group, the longest matching rule wins. That trips people up constantly: a User-agent: * block with Disallow: / does not necessarily stop a crawler that has its own named group elsewhere in the file.
Also worth remembering that robots.txt expresses intent. It is not access control. If you need to actually restrict access, that is a different mechanism.
Where the two overlap
Both systems reward pages that answer a question directly and can be understood out of context. Both punish pages that bury the answer. Both depend on the crawler being able to reach the page at all.
So the shared work is real: write pages that state their point, keep them crawlable, and make sure the crawler you care about is not blocked. The divergence is in what you do after that. On Google, you are still optimizing for ranking. On ChatGPT, you are optimizing for retrievability and quotability, and the ranking concept does not map cleanly.
If you want the ChatGPT side in more detail, how ChatGPT search picks the sources it cites covers the retrieval behavior.
Check your own setup before you optimize
The most useful hour you can spend this week is not a content project. It is an audit.
Open your robots.txt and answer three questions. Is Googlebot allowed on the pages you want in AI Overviews? Is OAI-SearchBot allowed, separately from GPTBot? And did you block Google-Extended thinking it controlled AI Overviews?
Then verify the outcome rather than assuming it. Fetch a page as Googlebot in Search Console to confirm it is indexable. For the ChatGPT side, check whether your pages show up when you ask ChatGPT search the questions your buyers ask. If you want to track that over time across ChatGPT, Gemini and Perplexity, that is what PromptRanking is built for, but a manual spot check today will tell you whether you have an access problem or a content problem. Fix the access problem first, because no amount of rewriting helps a page the crawler cannot reach.