For years, Public Relations has shaped what people read, believe and repeat. Now it’s doing the same thing for machines.

Anyone who works in PR already knows this. Where a story lands matters just as much as what it says. However, the audience has changed. It’s bigger. And quieter.

AI tools are fast becoming a front door to information. Google’s search market share has dipped below 90% for the first time since 2015, while daily AI usage has grown by 233% over the same period. In healthcare especially, we are seeing that people aren’t just Googling symptoms and clicking around anymore. More and more, they’re asking AI!

Why does this matter? Because AI doesn’t invent knowledge from scratch. It learns from what’s already out there. When the stakes are high, as they are in healthcare, it looks for signals of credibility. The New York Times has already flagged growing concerns around misinformation in AI health advice. A Guardian investigation uncovered that people were being put at risk of harm by false and misleading information, resulting in Google removing some of its AI health summaries. And with the recent launches of ChatGPT Health and Claude for Healthcare, the quality of the material feeding these systems is more important than ever.

What’s easy to overlook is how differently time works here. News moves fast. Coverage appears and disappears within days (sometimes less!). AI doesn’t behave like that. It tends to favour material that’s had time to settle. Articles that have been published, referenced and reinforced over months. Sometimes up to a year old. In other words, yesterday’s coverage doesn’t die. It lingers. It compounds. It gets a second life.

If you look at where AI pulls its sources from, the pattern is pretty stark. Roughly 89% of citations come from earned media. Trade press, specialist outlets, blogs, LinkedIn articles and newswire content. Not paid campaigns. Not ads. Not brand websites.

So PR isn’t just influencing human audiences anymore. It’s leaving notes in the margins of the algorithm.

Earned media: AI’s comfort read.  

Like most of us, AI has a type. It leans towards content that’s specialist, credible and keeps showing up over time. That’s great news for PR teams, especially in healthcare, where depth, accuracy and getting it right matter far more than shouting the loudest.

A big splashy headline still looks impressive in a boardroom. Sure. But it’s no longer the whole story. What matters just as much is the coverage that keeps resurfacing in AI-generated answers. Trade titles and professional blogs have long been treated as niche, sometimes quietly under-valued because their audiences look small on a media plan and lack household recognition. That mindset is starting to feel outdated.

When AI is being used by almost everyone, audience boundaries blur. Information from specialist publications doesn’t stay put. It gets re-served, re-contextualised and pulled into answers at scale. A piece published months earlier can suddenly become the main explanation of a topic. It can become gospel.

That changes how we should be thinking about media lists. They don’t need to stop at the usual names. They should include the outlets and voices AI keeps returning to. The value isn’t just who reads something on the day it’s published. It’s how that piece shapes the version of the story that gets repeated later.

When an AI tool answers a question about a brand, a therapy area or a health issue, it isn’t pulling from a single article. It’s stitching together what it has absorbed over time. Those sources don’t just decide what gets mentioned. They shape the tone and general emphasis. High-quality earned media doesn’t just land coverage in the moment. It trains the system that will go on to repeat it.

And once something is in there, it sticks around. Outdated narratives, inaccuracies or sloppy framing don’t simply fade away with the next news cycle. They can resurface long after the original article has been forgotten.  

In healthcare, that matters. When AI outputs can influence understanding, decision-making or behaviour, accuracy and balance aren’t nice-to-haves. We work within some of the most rigorous compliance frameworks in the industry, which means branded messaging is mostly reserved for trade press – already a favoured source for AI models. But it’s not just branded messaging that shapes what AI serves up. Unbranded disease awareness is just as important, as it’s this credible, accessible information that gets pulled into AI answers for the wider public, helping bridge the gap between professional content and everyday health decisions.

What we’re not measuring (yet)

PR has never had an easy relationship with measurement. Legacy metrics like Advertising Value Equivalency offered reassurance through simplicity but stripped away much of what makes PR valuable in the first place. Even today, familiar measures like Share of Voice only tell a small part of the story.

That gap is becoming harder to ignore. If PR activity is shaping what AI systems say, then relying on traditional measures alone feels increasingly incomplete.

The industry’s slowly catching on. Around 61% of brands are already trying to understand how they show up on AI platforms – whether that’s audiences going directly to tools like ChatGPT for answers or encountering AI-generated summaries and citations within search experiences such as Google’s AI Overviews. Two thirds of PR professionals expect AI citations to become a standard metric in the not so far future. This isn’t about chasing another shiny number. It’s about acknowledging a new layer of influence alongside the channels we already care about. Paid, social and owned media still matter. Earned media just has an additional job now.

Future-facing measurement needs to ask slightly different questions. How visible is a brand in AI-generated answers? Which sources keep being referenced over time? How consistently is the story being repeated? And are outdated or misleading frames still floating around in the system?

None of this means scrapping existing frameworks. It means adapting them. Tracking patterns over time. Anchoring everything back to real business objectives.

PR’s core strengths have always been visibility, credibility and trust. AI simply makes those strengths more concrete. Earned media doesn’t only shape perception in the moment. It has an afterlife. One that increasingly influences how people make sense of the world.