Large language models (LLMs) have transformed the way people search for and consume information. While traditional search engines like Google and Bing remain popular, nearly half of U.S. and European internet users now begin their search with some kind of AI tool. 

Even traditional search engines now utilize AI-generated summaries, sometimes leading to so-called “zero-click searches” as users accept responses at face value without verifying their sources. 

A new study by Fullintel, in partnership with Dr. Tyler Page of the University of Connecticut (UConn), shows that PR teams are uniquely positioned to capitalize on the growing use of AI search engines. That’s because the study demonstrates that nearly half of AI citations are from news and information sources – the types of sources PR pros pitch practically every day. 

The study was recently accepted into the competitive list of studies to be presented at the International Public Relations Research Conference (IPRRC) in March (with an average acceptance rate just over 50 percent).

How Large Language Models Find Information

Search is evolving rapidly. And while the basics of search engine optimization (SEO) are relatively well established, it’s less clear which factors make a particular piece of content – such as a news article, academic journal article, or corporate blog post – more likely to be cited in an AI search engine output.

AI platforms – which synthesize information and cite sources within generated responses – have already fundamentally altered how (and which) content reaches audiences. AI models, such as ChatGPT and Gemini, employ a technique known as retrieval-augmented generation (RAG) to locate and synthesize information by ranking and selecting content based on semantic similarity, clarity, and structural accessibility. 

Studies have suggested that AI retrieval tends to favor content that is:

• Well organized (through headings, bullets, tables, and other visual elements)
• Clearly sourced, with visible dates and author information.

But what does that mean in terms of which types of sources get cited most often, and how are those sources selected?

The answer is particularly consequential for health communications, where the accuracy and credibility of information can have major implications for health outcomes.

How to Get Cited in AI Engines? Pitch to Credible News And Info Sources

Through an analysis of weight loss drug-related content, the Fullintel-UConn study examines in depth how AI engines select and prioritize earned media.

The final study will analyze which article characteristics – including reach, credibility, sentiment, brand prominence, and voices – most often influence AI search engine citations and provide greater visibility for certain types of online media.

After inputting 400 distinct prompts across 10 personas reflecting a broad U.S. demographic distribution, the final content analysis found:

• Forty-eight percent of coverage originated from a combination of non weight-loss drug corporate websites, universities, and health networks. 
  • What it means for PR pros: These credible, often well-funded sources present a sizable opportunity for communications teams in terms of partnerships or other collaborations.
• Forty-seven percent originated from third-party news and other credible informational sources, such as association websites and other sources that aggregate and present content to inform their audience objectively. 
  • What it means for PR pros: The findings reiterate the value of PR teams pitching relevant story ideas and topics to credible news and information sources, which drove almost half of all AI citations. 
• The final five percent of citations were from a combination of what were classified as owned coverage from weight-loss brand websites or government websites.
  • What it means for PR pros: While showing up far less often in AI engines, these sites also present far less of an opportunity for proactive PR pitching. 

The top news sources included medicalnewstoday.com, healthline.com, and drugs.com. Examples of informational sources included diatribe.org, pewresearch.org, diabetes.org, and wikipedia.org. 

Articles appearing in news feeds on corporate and health network websites were categorized as news coverage, while other content appearing on these sites was classified as other third-party coverage. 

The analysis team examined earned media articles cited by Scrunch AI in response to queries about weight loss medications Ozempic, Mounjaro, Wegovy, Zepbound, Phentermine, Rybelsus, and Trulicity. Articles from news outlets were extracted and coded for multiple variables, including outlet authority, article structure, source credibility, sentiment, brand prominence, and social engagement.

Beyond SEO: Takeaways for PR professionals

As generative AI changes the way people search for and consume information, PR professionals must familiarize themselves with generative engine optimization (GEO) and identify the content characteristics that are most effective in driving citations from AI engines. 

The Fullinte study aims to equip PR professionals with a framework to optimize visibility and credibility in AI-generated search results. It shows that PR professionals are uniquely well-positioned to capitalize on the rapid growth of AI engines and AI search:

• Credible news and information websites accounted for nearly half of all AI engine citations in the study.
• Because PR teams regularly pitch these types of websites, they are well-positioned to become more valuable than marketers or SEO experts when it comes to generating AI visibility and citations for their brands. 
• PR pros can use these findings as an evidence-based framework to help tailor their content and outreach strategies, and enhance visibility in the age of AI search. 

The study will be presented at IPRRC’s 29th International PR Research Conference in Orlando, Florida, in March 2026.