Publishers have a unique relationship with AI. Your articles, guides, and reporting are part of the training data that models like ChatGPT, Claude, and Gemini learned from. Your content shaped what these systems know. But that doesn't mean they credit you when they use it.
The uncomfortable reality: AI systems trained on your work can confidently answer questions your journalism made possible, without ever mentioning your name. And the way different AI platforms handle attribution varies enormously.
Understanding that gap between contributing to AI's knowledge and being visible in AI's outputs is the first step to doing something about it.
The citation gap: training data vs attribution
When ChatGPT answers a question about, say, the health effects of ultra-processed food, it draws on thousands of articles it absorbed during training. Some of those articles were original reporting from health publications. Some were peer-reviewed research summaries written by science journalists. The model synthesizes all of that into a confident, fluent answer with no footnotes.
This is fundamentally different from how Google Search works. Google sends traffic to the sources it indexes. AI knowledge models consume those sources and produce answers that replace them entirely.
Perplexity and Google AI Overviews work differently. They search the live web and cite sources inline. But even here, citation doesn't always mean the original publisher gets credit. Perplexity may cite a secondary source that aggregated your reporting, or a Reddit thread that discussed your findings, instead of your original article.
The result is a two-tier system. AI Search tools (Perplexity, AI Overviews) can cite you but may not. AI Knowledge tools (ChatGPT, Claude, Gemini) used your work to learn but structurally cannot attribute it in most responses.
How Perplexity decides what to cite
Perplexity is the most transparent AI system when it comes to citations. It searches the web in real time, reads pages, and lists its sources. For publishers, this is the most direct path to AI visibility. If Perplexity can find your content and considers it authoritative, it will link to you.
What influences whether Perplexity cites your publication over a competitor's:
Recency. Perplexity prioritises fresh content. If you published the definitive guide to a topic last week, you're more likely to be cited than an older piece, even if the older piece is more comprehensive. This is one area where publishers have a structural advantage, since you produce new content constantly.
Topical authority. Publications that consistently cover a beat get cited more for queries in that domain. A technology publication with years of AI coverage is more likely to be cited for AI questions than a general-interest outlet that wrote one AI piece.
Structured, extractable content. Perplexity reads and synthesizes. Content that's well-structured (clear headings, definitive statements, data points that can be extracted) is easier for AI to cite accurately. Long, narrative-heavy pieces with buried conclusions are harder for AI to use.
Domain authority. This still matters. Established publications with strong link profiles and high domain authority tend to appear more frequently in Perplexity's citations, for the same reason they rank well in traditional search: they're trusted sources in the web's link graph.
How ChatGPT and Claude handle publisher content
Training-data models work differently. They don't cite sources per response. Instead, they synthesize everything they've learned into a single answer. For publishers, this creates a different kind of visibility question: does the model know your publication exists as an authoritative entity?
You can test this directly. Ask ChatGPT or Claude: "What does [your publication] say about [topic]?" or "Is [your publication] a reliable source on [topic]?" The response tells you whether the model recognizes your publication as a distinct, credible entity, which influences whether it draws on your work (even without explicit citation) when answering related queries.
Publications with strong entity recognition (a Wikipedia page, consistent coverage across the web, a clear topical identity) tend to be referenced by name more often. Models will say things like "according to sources including [publication]" or "publications like [publication] have reported that..." when they associate your brand with authoritative coverage of a topic.
Publications without that recognition get absorbed anonymously. Their reporting influences the model's knowledge, but the model doesn't distinguish their contribution from the general web.
What publishers can do
The levers for publisher AI visibility overlap with traditional SEO, but aren't identical to it.
Be the definitive source on your beat
AI systems, whether searching live or drawing on training data, favor content that provides clear, authoritative answers. If someone asks "what is [concept]?" and your publication has the most comprehensive, well-structured explainer, that's the page AI will cite or learn from.
This means investing in reference content, not just news coverage. Evergreen guides, definitive explainers, and data-driven analyses that remain useful over time give AI something to anchor to. News articles drive traffic but expire quickly. Reference content compounds in AI visibility.
Structure content for extraction
AI systems extract facts, not narratives. Content that buries its key findings in paragraph seven of a 2,000-word feature is less useful to AI than content that leads with clear findings and supports them with structured evidence.
This doesn't mean dumbing down your writing. It means structuring it so that the most important claims are surfaced clearly through headings, summary boxes, and clear topic sentences. Schema markup (Article schema, FAQ schema, author markup) helps AI systems categorize and attribute your content correctly.
Build entity recognition
Your publication needs to exist as a recognized entity in AI's knowledge base. This means:
- A current, accurate Wikipedia page (if your publication meets notability criteria)
- Consistent Wikidata entries
- A Google Knowledge Panel
- Consistent use of your publication's full name across all platforms
- Author pages with structured markup that link contributors to your publication
Entity recognition is what separates "a website that published some articles" from "a recognized publication that AI considers authoritative." It's the difference between your work being absorbed anonymously and being attributed by name.
Cultivate cross-platform presence
AI models learn from the entire web, not just your domain. When other publications cite your reporting, when your journalists are quoted on podcasts, when your analysis is referenced in academic papers or industry reports, all of that strengthens your publication's presence in training data.
Syndication, guest contributions, and being a source for other journalists don't just build traditional authority. They create the web-wide footprint that AI uses to assess who matters in a given topic.
Monitor what AI says about your publication
The most actionable step is knowing where you stand. What happens when someone asks Perplexity a question in your core coverage area? Do they cite you? When someone asks ChatGPT whether your publication is a reliable source, what does it say? When AI compares coverage of a topic, does your publication appear?
These are measurable questions, and the answers reveal exactly where your AI visibility is strong and where it's weak.
The publisher-specific opportunity
Publishers have an advantage that most brands don't: you create the content AI values most. Original reporting, expert analysis, and primary research are what AI needs to give good answers. The challenge isn't creating valuable content. It's ensuring that value is recognized and attributed.
The publishers who treat AI visibility as a strategic priority now, who measure it, understand the differences between platforms, and structure their content accordingly, will maintain their authority as AI becomes a primary information channel. The ones who assume that good journalism speaks for itself will find their work powering AI answers that never mention their name.
SeenByAI measures your visibility across all major AI platforms (Perplexity, ChatGPT, Claude, Gemini, and Google AI Overviews) and generates a prioritized action plan to close the gaps between where you are and where you should be. Get started free.
Written by Stu Miller, Founder of SeenByAI and CEO & Co-founder of Smart Insights. Stu has spent 16 years helping ecommerce businesses grow their digital marketing, and built SeenByAI after experiencing the AI visibility problem first-hand running his own Shopify store.
Related reading: