Earned Media Still Drives Generative AI Citations as Press Release Visibility Grows

New Muck Rack Report finds Press Release Citations Increased 5x since July 2025

Muck Rack, the company behind Generative Pulse, the tool that helps PR and communications teams define how their brands appear in AI-generated search results, released new findings from its What Is AI Reading? report, revealing that earned media continues to shape the information generative AI cites. The report analyzed more than one million links cited by leading AI models and outlines how citation patterns have shifted since the company first studied responses to realistic user prompts in July 2025.

The new findings show that non-paid media continues to dominate AI citation behavior. About 94% of all citations come from non-paid sources, and earned media alone accounts for 82%. Journalism remains a major influence, accounting for about 20-30% of citations over any period.

However, even knowing that generative AI relies on earned media, PR professionals aren’t pitching the right people. When comparing the most pitched journalists on Muck Rack with those most cited by AI engines for a particular brand, the overlap is only 2% on average.

Marketing Technology News: Martech Interview with Meena Ganesh, Senior Product Marketing Manager @ Box AI

“What we saw in our July report is even clearer now: earned media still shapes how AI understands brands. At the same time, the gap between who PR teams pitch and who AI cites is striking–only a 2% overlap, which shows the industry hasn’t fully adapted,” said Greg Galant, cofounder and CEO of Muck Rack. “This report gives communicators clear steps to close that gap, and those who act now will move their organizations forward while demonstrating their value through impactful metrics. That means more budget, more headcount and more opportunities for their teams.”

Press Release Citations Increased 
Press releases grew 5x since July 2025, which is driven by higher citation rates in ChatGPT and Gemini. The findings also show that structure plays a role in what AI models choose to cite. The majority of cited press releases include:

• 2x as many statistics in them on average
• 30% more action verbs
• 2.5x as many bullet points
• 30% higher rate of objective sentences
• More unique companies and products

Other key insights and trends from the What is AI Reading? December 2025 report:

• Recency shapes answers: Half of all citations come from articles published within the last 11 months. About 4% come from the prior week.
• Citations continue to shift across models: Since July, AI models have decreased reliance on Wikipedia. They also reduced citations from large consulting firms, driving a decrease in citations from third-party corporate and blogs from 37% to 24% during the same period.
• Most cited sources vary: Top-cited outlets also shifted across models, with minimal overlap. Among the most frequently cited sources were: U.S. News and World Report, Nature, Yahoo Finance (Claude), Reuters, The Verge, The Guardian (ChatGPT) and Forbes, Investopedia, NerdWallet (Gemini).
• Owned content helps fact-finding: Owned content is most influential when users ask direct, fact-based questions about a brand. For discovery questions, AI models rely more on earned media and journalism.

Muck Rack’s What Is AI Reading? report is powered by Generative Pulse, which helps communications teams monitor and shape brand visibility inside AI engines. The full report includes detailed findings for ChatGPT, Claude and Gemini, along with industry-specific insights.

Marketing Technology News: Feature-Rich to Functionally Effective: Adjusting your Martech Strategy

Write in to editor@pressreleasecc.com to learn more about our exclusive editorial packages and programs.