Muck Rack Study: Generative AI Relies Heavily on Earned Media and Journalism

Miami, July 23, 2025 (GLOBE NEWSWIRE) — In a groundbreaking study analyzing millions of AI-cited links from hundreds of thousands of prompts, Muck Rack, the leading provider of the award-winning PR software, has revealed how generative AI prioritizes which sources to cite, and why this shift is critical for communicators looking to shape brand perception, influence buyers, and lead in the future of marketing.

The research, titled “What is AI Reading?” set out to answer a fundamental question: Does media coverage materially affect what AI says? The answer, unequivocally, is yes. 

“This study is an eye-opener for PR and communications teams aiming to understand how media mentions influence AI-generated content,” said Greg Galant, cofounder and CEO of  Muck Rack. “Until now, we’ve had theories and early signals—but now we’ve got solid evidence that earned media directly influences AI-generated output. This changes the stakes for PR. The way businesses are represented by AI now ties directly to the media coverage they earn.”

Study Findings

Citations actively influence AI output: Controlled prompt testing showed that when citations are enabled, LLM (large language models like OpenAI’s ChatGPT and Google Gemini) outputs meaningfully change. Earned media doesn’t just show up—it affects what is said. This proves that cited content is not decorative. It materially grounds outputs in real-time, dynamic inputs. 

Earned media is a foundational input: More than 95% of citations come from unpaid media sources and 85% of those come from earned sources, while another quarter are from journalistic sources. Half of total AI responses included at least one earned media citation. These figures highlight the critical role that earned media strategies play in GEO (Generative Engine Optimization) and AIO (Artificial Intelligence Optimization) visibility.

What Determines Whether a Brand is Cited

Three key variables drive citation inclusion:

• Recency: Particularly in OpenAI models, fresh content–especially on topical, opinion-based, or event-driven queries–is prioritized.
• Query framing: Advice-seeking or opinion-based prompts trigger more dynamic citations, while encyclopedic queries tend to fall back on older, static training data.
• Outlet authority: High-domain authority outlets such as Reuters, Axios, Financial Times, AP, TIME, Forbes, NPR and CNN are frequently cited—but even among these, performance varies. In addition, more niche sites, such as Good Housekeeping and Investopedia, are also frequently referenced.

Strategic Implications for Communications Teams

Communications strategies must evolve to account for this new layer of visibility:

• New KPIs: Success now includes whether your brand is cited in AI responses—not just traditional media placements or backlinks.
• New tactics: PR teams should prioritize high-authority outlets and align publishing cadence with the recency preferences of generative AI models.

As AI increasingly mediates how information is accessed and consumed, it’s clear that PR strategy must expand beyond human readers to include algorithmic interpreters. The study confirms what many in the industry have suspected: the stories you place—and where and when you place them—can now shape AI understanding at scale.

Galant added, “Generative AI is reshaping itself at an unprecedented pace, with new patterns emerging constantly, so the brands that thrive will be those that proactively track their position and adapt in real time to stay ahead.”

To access the full report, visit generativepulse.ai/whatisaireading.

Methodology

This study explores how modern generative AI systems cite sources in response to realistic user prompts. The objective was to quantify and characterize the nature of AI-generated citations across different use cases and vendor models. This includes their frequency, source types, and the prominence of earned and owned media.

To accomplish this, Generative Pulse and Muck Rack constructed a large, diverse prompt set and executed it across several web-enabled language models, followed by systematic analysis of the responses and the cited links. The prompts span all industries and subject matter. Sometimes they specifically mention companies by name, sometimes they do not.

The following specific models were used to execute hundreds of thousands of queries, during the month of July 2025:Chat GPT (both ‘4o’ and ‘4o-mini’), Gemini (‘flash’ and ‘pro’) and Claude (‘sonnet’ and ‘haiku’).

# # #

About Generative Pulse

Generative Pulse is a tool that helps PR and communications teams monitor and define how their brands appear in AI-generated search results. Built around the concept of Generative Engine Optimization (GEO), it reveals which journalists, outlets, and sources influence large language models like ChatGPT—so teams can shape brand visibility in the age of AI. Integrated into Muck Rack’s PR platform and backed by a $180M Series A financing, Generative Pulse gives comms professionals the insights and tools to lead in an AI-native search landscape. Learn more at generativepulse.ai.  

About Muck Rack

Muck Rack is the leading provider of award-winning PR software built for how brands and agencies work. The only public relations software powered by intuitive technology and the most accurate, comprehensive data provided by journalists themselves, Muck Rack combines media database, monitoring and reporting for seamless team collaboration, pitching and measurement. Purpose-built for communications and public relations, Muck Rack helps more than 5,000 companies worldwide analyze and report on the impact of their media relations. Thousands of journalists use Muck Rack’s free tools to showcase their portfolios, analyze news about any topic and measure the impact of their stories. Learn more at muckrack.com.

We’re Hiring!

Wall St Business News, Latest and Up-to-date Business Stories from Newsmakers of Tomorrow