What is AI Reading? The Muck Rack AI Report 2025

Summary

This post breaks down findings from Muck Rack’s AI report, “What is AI Reading?” on how AI search engines decide which sources to trust and cite. The key takeaway is that earned media, not owned or paid content, drives AI visibility. Recency, source credibility, and question style all influence whether a brand appears in AI-generated answers. Different models like ChatGPT, Claude, and Gemini cite different sources, making it essential to monitor each one. For PR teams, this means shifting from splashy campaigns to a steady stream of high-authority coverage. The real risk is staying invisible while competitors feed the models with stronger signals.

AI search is rewriting the rules of visibility. Muck Rack’s What is AI Reading 2025 report spells out exactly how it works. It doesn’t just tell you which sources AI prefers. It shows how those sources shift based on the question, the topic, and the model. If you want your brand to show up in generative AI results, you need to stop thinking like an SEO. Start thinking like a journalist.

Let’s break it down.

Citations Shape the Narrative

Turn citations on, and the AI response changes. Not just the source list. The entire answer. Muck Rack tested this across a million links and found that enabling citations doesn’t simply add proof. It rewires the response itself, and it should always be a best practice.

AI relies heavily on its source material. If that material includes your coverage, your story gets baked into the response. If it doesn’t, your brand disappears.

STRATEGIC INSIGHTS

Most brands obsess over SEO and miss the bigger threat. AI search doesn’t care about optimization tricks. It pulls from what it sees as credible, and that starts with independent coverage. If your comms strategy doesn’t include a steady flow of earned media, your brand will fade from AI-generated visibility. The real blind spot is assuming past coverage is enough. AI favors the now, not the archive.

Earned Media Gets Picked First

Over 95% of links cited by AI come from non-paid sources. Journalism leads the pack, especially in recent, fast-moving topics. For queries involving recency, nearly half of all citations are journalistic.

Generative AI citations come from earned media: Muck Rack GEO

Owned media and paid content barely register. And when they do, it’s often buried behind more credible, third-party links.

STRATEGIC INSIGHTS

Most PR teams focus on generating owned content without realizing how little weight it carries in AI search. The hidden risk is assuming that owned assets like blogs or press releases will surface when they rarely get cited. AI models are tuned to prioritize third-party validation, not branded messaging. Executives need to ask a different question: Where is our earned media showing up, and are those sources authoritative enough to feed the AI citation loop? The real issue isn’t visibility. It’s the erosion of narrative control when brands rely on channels AI deprioritizes.

The Prompt Matters More Than You Think

AI doesn’t just respond differently based on the model. It shifts its citations based on how the question is asked.

Fact-based prompts pull from encyclopedic or static sources. Advice-seeking or opinion-based prompts trigger more diverse and dynamic responses. Claude, for example, cited Reuters far less often than ChatGPT or Gemini.

STRATEGIC INSIGHTS

Most executives overlook how AI filters content based on prompt style. If your message only fits one type of query, it disappears in others. The hidden risk is that AI models prioritize different content depending on how people ask questions. This fractures visibility. A press mention that ranks high in one scenario may never appear in another. PR leaders should audit not just where they appear, but how often they show up across different prompt categories. That is where narrative control is truly won or lost.

Authority Is Relative, Industry Matters

The usual heavyweights like Reuters, FT, and AP get cited often. But they don’t dominate across every sector. AI favors niche sources when the query is industry-specific.

In Energy, for example, government and research institutions climb the ranks. In Hospitality, owned media holds unusual influence. Finance leans into consumer advice sites. Context drives credibility.

STRATEGIC INSIGHTS

The real risk here is assuming your most prestigious placements are your most valuable. They’re not. AI weighs relevance more than reputation, which flips traditional PR logic on its head. A well-placed mention in a niche trade journal may do more for your AI visibility than a quote in The Wall Street Journal. Most execs don’t ask how coverage ranks within industry-specific AI citations. They track top-tier hits, not AI traction. That’s a missed opportunity hiding in plain sight.

Recency Drives Relevance

Fresh content gets rewarded, especially by OpenAI models. Over half of the journalism links cited by ChatGPT were published within the past year. For Claude, that figure drops to 36 percent. Still meaningful, but less aggressive.

Of course, this is great news for brands that have news to share. Recency drives relevance, but it should be newsworthy and add value at the same time.

STRATEGIC INSIGHTS

Most PR teams think in campaign cycles. But AI models don’t. They weigh recency with no regard for narrative arc or strategic timing. That creates a blind spot. A one-time media blitz won’t sustain AI visibility. AI platforms prefer a drip of relevant, recent coverage over a splashy launch that fades. This forces a shift in strategy. Your comms plan should mimic a newsroom cadence, not a product roadmap. The real insight? Visibility isn’t driven by volume. It’s driven by sustained relevance over time.

Different Models, Different Behaviors

Claude, Gemini, and ChatGPT do not act the same. Claude tends to ignore journalism more often. Gemini leans into financial and government sources. ChatGPT offers the most balanced mix but still prioritizes journalistic authority.

This is good news if you are in charge of PR or media relations.

STRATEGIC INSIGHTS

PR teams often rely on screenshots or single-platform tests to understand AI visibility. That’s a dangerous shortcut. Different models pull from different sources, rank authority differently, and cite entirely separate domains depending on the topic. The hidden risk isn’t just underreporting. It’s misdiagnosing your brand’s exposure. A flattering result from ChatGPT could mask poor visibility in Gemini or Claude. If your listening strategy doesn’t compare performance across models by industry and topic, you’re tracking a false signal.

MUCK RACK GEO: What This Means for PR Teams

The findings from the MuckRack AI Report and the What is AI Reading 2025 study make one point clear. AI search is not a future trend. It is a present-day battleground for visibility and reputation. Muck Rack shows how citations, prompt structure, and industry-specific authority change the way AI answers questions. These shifts create both risks and opportunities for PR leaders.

The connection to Muck Rack GEO is direct. Generative Engine Optimization is how you translate these insights into practice. It forces PR teams to shift their mindset from search marketers to journalists. The priority becomes credible, timely, and authoritative earned media that AI cannot ignore.

This is where PR strategy changes. Owned content and SEO tricks may give you a small bump, but they will not control how AI explains your brand. What matters is whether your earned media shows up in the sources AI prefers and whether your coverage is recent and relevant enough to be cited.

The next wave of competitive advantage in communications will come from how well teams adapt to these shifts. That means tracking AI visibility across models, auditing the sources that feed citations, and aligning your outreach with the cadence of newsrooms instead of campaigns. If you want to control your narrative in the AI era, you need to think like a reporter and measure like an analyst. PR is uniquely positioned to do both.