In AI answers about brands, the most frequently cited sources are earned publications: editorial articles, interviews, analyses and reports. According to Muck Rack's "What Is AI Reading?" report from May 2026, earned media, meaning coverage the brand did not pay for, accounts for 84% of citations in ChatGPT, Claude and Gemini. It is earned media that most shapes what AI models say about a brand.
Which media and publications does AI cite most often?
AI models build their answers from content they consider credible and useful. In practice, this means a strong tilt toward earned publications: material prepared by independent journalists and newsrooms, experts and the authors of industry reports. In the Muck Rack study, the dominance of earned media has held at a similar level across all three editions of the report since July 2025. That one fact explains why traditional media relations are not going away but are gaining a new dimension in the AI era.
The most weight goes to editorial articles about the brand, its offer, its implementations and the results of its work, as long as they have clear attribution, quotes and a description that lets AI grasp the meaning and context. Expert interviews and comments play a similar role, especially those that explain how a product works, show know-how and tie the brand to a specific category. Reports, analyses and studies give AI a strong signal about a brand's relevance to a topic, provided the brand is mentioned in them. Case studies and project write-ups in the media, where the newsroom describes the results of the brand's work and names its sources, supply material for building a narrative.
A brand's own materials can also work for visibility, especially when earned media cite them and AI sees a consistent chain of attribution: the newsroom links to the brand's site, and the brand keeps substantive, up-to-date content there. Still, in light of the Muck Rack data, it is earned media that is cited by AI most often.
How do publications affect a brand's image in the AI era?
AI answers user questions by combining information from many sources. If, in earned publications, the brand appears as the author of solutions, a proven partner or a quoted expert, that is the picture that has a chance of showing up in AI answers. Missing or superficial mentions, on the other hand, leave the models with nothing to "pull," and the brand tends to be skipped on the questions that matter to users.
Working with earned publications shows a few patterns. The more clearly the brand name appears in the text and the more explicitly a specific solution or competence is attributed to it, the easier it is for AI to link the brand to the topic. When the same claim about a brand repeats across many publications, AI is more willing to treat it as credible, stable information. Models make better use of content that clearly describes results, mechanics and context, while overly general material dilutes the message. Publications that cite their sources and experts precisely are also easier to interpret.
Since earned publications dominate citations, a brand's reputation in AI depends directly on whether and how independent newsrooms describe it, and on which studies and reports mention it. If a brand does not supply content that can be quoted, AI has a reason to point to other players.
Monitoring brand visibility in AI answers is now a daily PR practice. Without it, you do not know whether earned publications are really working for your reputation. Your strategy should combine consistent media relations with optimizing content for citability by AI – from clear attribution to a plain description of the value the brand brings to a topic.
Which PR strategies can increase brand visibility in AI answers?
Media relations today means working on two fronts: you have to convince the journalist and, at the same time, take care of how AI models will read and use the publication. Start by defining the questions you want AI to answer with your brand. These are everyday user questions like "who solves problem X," "how to choose Y," "examples of Z." If you do not know which questions your customers ask, you do not know whether AI has any reason to mention you.
On that basis, develop clear, quotable claims that explain what the brand does and why it is worth recommending, along with evidence: statements from an expert, descriptions of implementations, educational material and content the media like to refer to. When planning earned publications, avoid generalities. Use the brand's full name in key paragraphs, describe use cases and results precisely, and point to sources, so that a newsroom can easily find a passage in the text that it can cite.
A simple checklist helps when working with newsrooms:
- Instead of asking for a "mention," offer content that adds value to the article: explanatory boxes, definitions, checkpoints, step-by-step tips.
- Check whether your site has substantive content that develops the topic on which the newsroom quotes you – AI checks whether the attribution is backed up.
- Use clear topical headings, definitions and bullet points so a model can match a passage to a user's question.
- Keep your spokesperson's or expert's statements consistent and repeatable on the topics you want to "own."
The goal is one: give AI material that is easy to cite, set in a credible context and consistent in its attribution to the brand. Content like this most often "passes" the models' filter and makes it into answers.
How do you monitor the results of PR in the context of AI?
Without systematic measurement, you cannot verify whether earned publications really influence what AI says. A measurement model should answer two questions: is the brand mentioned in AI answers to key questions, and which sources does AI draw its knowledge from.
The basic metric is brand visibility in AI, that is, whether the brand appears in answers to defined questions and whether AI treats it as a relevant answer in a given category. Share of voice shows how often other brands appear in the same questions and which ones, so who you are most often compared with. Blind spots, meaning questions where AI mentions competitors and not your brand, are the simplest way to prioritize topics for PR and content work. The list of sources cited by AI tells you which newsrooms and which types of material "work" hardest for you.
Add to that alerts about significant changes in answers, such as a new competitor on a key question or a brand mention disappearing, and a report for the board that explains the relationship between publications and AI visibility and sets priorities for the coming weeks.
This kind of monitoring is done with a tool that regularly asks a set of questions in the main engines and shows whether the brand was mentioned, who AI recommends alongside it, which sources it draws on and what is worth changing. One example is Attent (getattent.com), which every week asks questions in ChatGPT, Gemini, Perplexity and Google AI (AI Overviews) and reports, among other things, the brand's AI visibility score, share of voice against competitors, blind spots, sources cited by AI and change alerts, and also provides a board report in PDF and a Google Search Console integration. Attent has analyzed millions of AI answers, and 78 companies use the tool.
Also set a steady working rhythm: how often results are reviewed, who owns which tasks, and a clear path from insight to publication. A simple flow of information between PR, SEO and product lets you refine claims faster, prepare an expert comment and close out a publication with the newsroom. A process like this keeps attribution consistent and raises the chance that AI will more often associate your brand with the target category in specific user questions.
What are the most important PR trends in the AI era?
The strongest direction is the role of earned media as the main "carrier." According to Muck Rack's 2026 report, earned media accounts for 84% of citations in ChatGPT, Claude and Gemini, so PR works even more on "content" that newsrooms can quote and AI can then use. This means working on two fronts – convincing the journalist and taking care of brand visibility in AI models – which changes how experts are briefed, topics are planned and materials are built.
Citability is becoming a content quality criterion. How materials are written and edited matters as much as their substance, because clear attribution, definitions, bullet points and "boxes" make it easier for AI to use the content. At the same time, monitoring AI answers is becoming routine: just as media monitoring became a standard, tracking whether and how the brand is mentioned by ChatGPT, Gemini, Claude, Perplexity and Google AI (AI Overviews), and which publications stand behind the answers, is now a fixture.
Collaboration with newsrooms is changing too. Brands more often offer material that is useful to both readers and AI models, that is, material with clear explanations, examples of use and consistent attribution. All these trends lead to one conclusion: the better you plan earned publications and how they are reflected in AI answers, the more stable and predictable a narrative about the brand you will build.
Examples of using brand visibility monitoring tools
The best way to start working with a tool that monitors AI answers is with a diagnosis. Take a list of key user questions and check whether the brand is mentioned in AI answers and who it is compared with. Turn blind spots into a plan for earned publications by choosing topics where AI currently points to others.
The list of sources cited by AI is useful in editorial strategy: it shows which type of material most often works for your visibility – expert interviews, in-depth analyses or reports – and lets you ask a newsroom for the format that best supports citability. While you are at it, review the landing pages that publications link to and make sure the brand name, the description of the solution and the key claims are stated clearly in the first paragraphs, so that AI easily recognizes the brand in context.
When an alert shows a drop in visibility or a new competitor on a specific question, plan a quick content top-up: an expert conversation with a newsroom, a comment on a trend or an update of the source material. In board reports, translate monitoring into the language of decisions – which topics are worth strengthening with earned publications, where the investment brings a measurable increase in presence in AI answers and which newsrooms are the most "carrying" for the brand. Monitoring results also work well in team workshops, as a basis for working together on citable language and on repeating key claims in expert statements and press materials.
A tool that automatically collects data from the main engines, points out blind spots and sources cited by AI, and builds a board report shortens the path from diagnosis to action. It organizes the work, makes the process consistent and lets you turn monitoring results into earned publications faster.
We check this for you, every week, in four engines
In a 30-minute call, we'll show you how AI answers your customers' questions today and who it recommends instead of you.
Book a callFrequently asked questions
What are the 4 models of PR?
The literature most often cites four PR models described by James Grunig and Todd Hunt: the press agentry (publicity) model, the public information model, the two-way asymmetric model and the two-way symmetric model. Whatever the terminology, in the AI era the key is to combine work with newsrooms (earned publications) with care for how AI models read content. It is work "on two fronts" – toward the journalist and toward AI models.
What are examples of PR activities?
Activities that work well in the context of AI are earned publications in the media – interviews, expert comments, analyses and reports in which the brand is clearly attributed. AI models cite this kind of content most often: according to Muck Rack's 2026 report, earned media accounts for 84% of citations in ChatGPT, Claude and Gemini.
What is most important in PR?
The priority is influence over what audiences and AI models actually see and cite. Earned publications in the media have the greatest say in what AI models say about a brand. Effective media relations combine convincing the journalist with care for brand visibility in AI models.
What are the 4 main areas of activity in public relations?
PR areas of activity are defined in different ways – the most commonly listed are media relations, internal communications, crisis communications and community relations (for example, public affairs and corporate social responsibility). In the AI era, each of them is worth running in a way that produces earned publications with clear attribution to the brand – these are the ones AI models cite most often.
Which media are most important in PR?
Earned publications have the greatest impact. According to Muck Rack's 2026 "What Is AI Reading?" report, earned media, meaning material the brand did not pay for, accounts for 84% of citations in ChatGPT, Claude and Gemini. This means that editorial articles, interviews, analyses and reports are most often what AI bases its answers about brands and categories on.
What are the latest trends in PR?
The growing role of earned publications in what AI models say, work "on two fronts" – toward journalists and toward AI models – and regular monitoring of AI answers. In practice, this means creating content that newsrooms can cite and that AI can easily use, with clear attribution to the brand.
