What is a citation in AI answers? Definition
On this page
A citation in an AI answer is a link or reference showing which source a statement came from, so the reader can check it and read further.
How it works
Citations show up in two places that matter to website owners.
In public answer engines. Perplexity says every answer "includes numbered citations linking to the original sources" (Perplexity). Google's AI Overviews and AI Mode show "links to supporting websites" next to the generated text (Google Search Central). Being one of those links is the main way a site gets traffic from AI answers.
In your own assistant. A website assistant built on retrieval-augmented generationRetrieval-augmented generation (RAG) is a technique where an AI system first looks up relevant documents, then writes its answer using what it found. knows which passages it was given, so it can link the pages behind each answer.
There are two ways to produce citations. The simple one is to ask the model in its instructions to cite sources; this works most of the time, but the model can cite the wrong passage or invent a reference. The more reliable one is built into some model APIs. Anthropic's citations feature returns "the exact passages that support each claim", and because the API parses them, citations "are guaranteed to contain valid pointers to the provided documents" (Claude docs).
A valid pointer is not the same as a correct claim. The passage exists, but the sentence built on it can still overstate it, so careful systems check that the cited text actually supports the statement.
Why it matters for your website
For visitors, a citation turns "trust me" into "here is the page". Someone asking about contract terms or security can open the source and read the exact wording.
For you, citations are a feedback loop. When an assistant gives a wrong answer and cites a page, you know which page to fix. When it answers without a citation, the answer came from somewhere you cannot see, which is a warning sign; see groundingGrounding means tying an AI model's answer to specific, verifiable sources, such as your own pages or a search index, so each claim can be checked. and hallucinationAn AI hallucination is a fluent, confident answer from a language model that is false or unsupported, such as an invented price, policy or citation..
For visibility in public AI answers, cited pages tend to state a fact plainly, in text, close to a heading that matches the question.
Example
A visitor asks a B2B site's assistant about data location:
Q: Where is customer data stored?
A: Customer data is stored in the EU, in Frankfurt, with backups in Dublin [1].
You can request a copy of the data processing agreement from the security page [2].
[1] Security and data residency - acme.example/security
[2] Legal documents - acme.example/legalEach claim has a numbered source, and each source is a page the visitor can open.
Related terms
- Grounding: tying every answer to a source.
- Hallucination: what citations help catch.
- Answer engine: tools whose answers come with citations.
- AI Overview: Google's cited summaries in Search.
- Retrieval: the step that finds the passages to cite.
- Retrieval-augmented generation (RAG): the pattern that makes citations possible.
Sources
- Citations (Claude documentation) — accessed October 21, 2026
- How does Perplexity work? (Perplexity Help Center) — accessed October 21, 2026
- AI features and your website (Google Search Central) — accessed October 21, 2026