AI is changing how U.S. adults search for and consume health information.
It may also change pharma’s relationship with public relations.
The KFF Tracking Poll on Health Information and Trust found 32% of respondents had used an AI tool for health information in the past year, and a West Health–Gallup survey cited more than 66 million U.S. adults using AI for physical or mental health advice. A SingleCare survey conducted in March found even higher usage: nearly half (46%) of U.S. adults have used AI tools to answer questions related to medications; 51% look up what a medication is and what it’s for and 44% ask about side effects or dosage.
How AI engines are changing pharma’s control over messaging
Here’s why this is tricky for pharma, and why PR can provide some guidance.
When a patient asks a chatbot whether a newly prescribed drug is safe, the manufacturer doesn’t get to decide which source materials are used.
AI models are not equal opportunists when answering a query. They tend to favor sources that are neutral, evidence-based and institutionally credible. They aren’t listening to what a brand says about its drugs in its own channels.
An Outcomes Rocket study of 5,472 health citations from ChatGPT, Claude, Gemini and Perplexity found that 30.7% came from health media sites such as Mayo Clinic and Cleveland Clinic, while only about 23% came from primary research. In a Fullintel and University of Connecticut study of weight-loss drug queries, third-party news and informational sources made up 47% of cited links.
This means a pharma brand’s website, TV spots and ad budget carry less weight in these answers than what health publishers, trade reporters, institutions and patient groups say about it.
Those are the audiences PR work with.
For pharma and nearly every other industry, reputation will now be mostly written by others, which makes earned media the channel that counts.
Meanwhile, regulators are squeezing pharma’s most familiar channel, TV ads. The crackdown began on September 9, 2025, with announcements from the White House, HHS and FDA and over 60 warning and untitled letters targeting DTC ads; OPDP was on track to exceed 50 letters in 2026, compared to only five in all of 2024.
So the paid, brand-controlled channel is under more pressure just as AI search arrives.
Managing reputation and trust in the AI era
Reputation management is tricky to balance in the AI era.
Ironically, even though people use AI, trust in the answers engines provide is low: only 18% of chatbot health users rate the responses as very or extremely accurate. Their instincts are on target: 83% of providers reported having to correct information patients got from AI tools.
Worldcom’s 2026 Healthcare Monitor notes that the first few companies named in a generative answer carry outsized weight. The report concludes: “Early mention can strengthen brand recognition, perceived relevance, and reputation. A company mentioned second or third can exert more influence than a company mentioned later, even if both are present.”
Models draw on years of coverage. So a positive story, when well documented by credible sources, will resurface.
But pricing fights, mishandled recalls and inaccurate information also continue to resurface in answers long after the news cycle moves on.
Inaccurate information can flood an AI search. A BMJ Open audit found chatbots consistently responded with high confidence, rarely refusing to answer even when questions implied unsafe or unproven treatments. If an AI misstates a drug’s risks, the brand absorbs the fallout without having said anything.
On October 27, Tech Crunch reported about the death of Bowen Wan, a 31-year-old Amazon software engineer with epilepsy, who was told by the OpenAI chatbot to taper off her seizure medication and hide it from her doctors. Wan was using GPT-4o, which OpenAI has since retired.
Pharma can no longer rely on controlled channels to ensure accuracy, awareness and credibility.
What pharma communicators should do now
- Find out what the engines are saying. Run the questions patients, physicians and payers ask through AI engines, including ChatGPT, Gemini, Perplexity, Claude and Google’s AI Overviews. Record which companies are named, how they are framed, whether safety information is accurate and which sources are cited. Repeat regularly, because answers change.
- Earn coverage where the models look. For consumer health questions, that means health publishers and institutional sources. For industry and policy questions, it means trade outlets such as STAT, Fierce Healthcare, Modern Healthcare and Health Affairs. A placement there now shapes AI answers and human readers alike.
- Build relationships with the validators. Clinicians, academic medical centers and patient groups are the sources AI trusts most in health. Long-term partnerships with them do more than any single campaign.
- Publish facts a model can cite. Easy-to-understand content is less likely to be paraphrased inaccurately. So before release, ensure disease education is in plain language, safety information is clear and clinical trial results are accessible, and of course reviewed by medical, legal and regulatory teams.
- Treat pricing and access as reputation issues. Pricing is a hot-topic button. So is access. The record shows they undo goodwill faster than anything else and AI keeps that record in circulation.
- Watch for errors and safety signals. When you find inaccurate information, trace it back to the source. Then provide it with the accurate source material. This will not be a one- and-done fix. You’ll need to keep checking what the AI engines are saying and keep sharing the accurate information until it takes hold. Agree in advance on the team or person who monitors and responds.
Pharma’s traditional approach to external communications assumed people formed their impressions through the company’s own channels. But AI engines build answers from what credible third parties have published, often over years, and rarely from what a brand says about itself. In our work with clients on AI visibility, we’ve seen how quickly a company’s reputation becomes whatever the record says. And the business impact.
If you want to know what AI engines are saying about your brand today, that’s where our Audit*E work starts. Our newest version scans 11 AI platforms, including OpenAI, Anthropic, Google, Microsoft, xAI, DeepSeek, Moonshot, Meta, Mistral, Cohere and Perplexity, to learn what AI says about your brand. But we don’t stop there. Every problem it finds comes with a recommendation, and every recommendation is tied to a task. It will also share how you stack up against competitors.
