Pharma vs AI: Tackling the threat of misinformation

Healthcare professionals (HCPs) are being asked to do much more with much less time, under increasing pressure to access medicines information quickly, driving them to rapid answers from AI tools. 

In this recent webinar hosted by PharmaTimes, our expert panel shared valuable insights into how AI is helping or hindering healthcare, and how Pharma can respond to this new pharmacy environment.

Our panellists: Ana da Silva (Boots UK), Connor Thompson-Poole (University Hospitals Sussex), Robert Nobrega (Cority), Sanna Zaffar (Government of Jersey), 

Moderated by: Tracy Olla (Datapharm)

Risks of Using AI Tools for Medicines Information: Fabricated references and frequency of misinformation

Multiple speakers described AI tools generating non‑existent citations. Connor Thompson-Poole noted: “I’ve read many university submissions… hallucinating references that don’t really exist.”

A Nature analysis found that 2.6% of articles submitted to a computer science conference in 2025 contained hallucinated citations, up from 0.3% in 2024. Robert Nobrega also described the ‘Bixonimania’ hoax, where prank papers were picked up by AI and then “cited by humans and in peer‑reviewed literature.” These instances demonstrate how AI errors can propagate into real‑world clinical or academic decision‑making.

Robert highlighted that you automatically get an AI overview when searching online, and that over one billion health‑related searches use AI‑powered tools daily.  Thus, he concludes, pharmacists and patients may even be relying on AI without realising it.

Poll results also confirmed widespread exposure to incorrect AI outputs as 65% of webinar participants reported encountering incorrect AI information “occasionally,” and 15% “frequently.”

Challenges for pharmacists accessing medicines information

Even if all prescribers can spot errors, they may not have the capacity, and with such pressure on workloads, convenience may override caution - do all healthcare professionals have the time to check if the generative AI is hallucinating?

Time pressure and workload bring into focus the lengths pharma must go to

Pharmacists often need rapid answers in high‑pressure environments. Take, for example, a medicine being left out of a refrigerator which has lost power, and needing quick information on storage temperatures, or simply, an impatient patient.  When SmPCs or trusted sources don’t contain the answer, pharmacists may need to contact companies, which “may take some time to get back to me”, or else turn to quick answers from AI tools.

This highlights a core challenge: traditional medicines information content is not always fast, searchable, or complete enough for real‑time clinical decision‑making

Digital literacy and system variation across settings…and generations

Hospital and community practice differ widely. As Educator, Sanna Zaffar, noted: “it really depends on the digital literacy of the team… what kind of electronic systems are they already using? Are they paper‑based?” This inconsistency affects how reliably pharmacists can access trusted information. Moreover, Pharma should be educating HCPs on trusted resources and ensuring their information is made available there within pharmacists’ information-seeking journeys.

Ana pitched in on her experience in community pharmacy, pointing out that Boots pharmacists “don’t use it much” for actual answers about medicines. In their organisation it’s limited to “guiding them where to go” – for example, a trusted medicines information resource such as emc or the BNF. They may use it to rephrase their answers to patients so that they can explain a condition in more patient-friendly language.

Sanna expressed concern that newer pharmacists may not yet have the experience to spot errors: “With more junior pharmacists, they might take things more matter of fact… if it says it is, then they'll take that and run with it.” She highlighted the critical skill of discernment, which often comes with experience.

Opportunities for Pharma to optimise provision of medicines information content

The discussion repeatedly pointed to gaps that pharma is well‑placed to fill, as the volatility of AI‑generated information underscores the need for high‑quality, structured medicines data that AI tools can safely consume.

Pharma can help stabilise this by supplying structured SmPC data, standardised formats that AI tools can ingest safely and metadata that supports searchability and decision support.

Additionally, given concerns about a new generation of pharmacists and their propensity to use AI, pharma could contribute by producing training materials, offering evidence‑based summaries and highlighting how to quickly verify information against authoritative sources.

Speaking to patients in their own language

With sufficient cognitive load during the working day, switching to ‘patient language’ can also be a challenge. Pharmacists need fast, unambiguous answers, and need the tools to help explain a complex side effect in a patient‑friendly way. There was also a consensus that highlighted safety information specifically relevant to the patient, could go a long way to engaging them in their medication.

In a more literal sense, many patients simply won’t engage because the safety information isn’t in their native language. In the most recent census, around 820,000 people are recorded as saying they cannot speak English well or at all. Sanna suggested that the need to provide patient safety information in other languages was evident in practice.

The picture points to structured data, and more considered ways of delivering highly regulated information

The conversation paints a clear picture. Pharmacists are under pressure, short on time, and often working with incomplete or hard‑to‑access medicines information. Naturally AI tools are being increasingly used, sometimes unknowingly, but many of the mainstream tools pose real risks due to hallucinations, fabricated citations, and inconsistent accuracy.

Pharma has a significant opportunity to step in with better‑structured, clearer, more accessible, and AI‑ready medicines information that supports safe prescribing and reduces operational burden. It can do this by delivering information used more reliably by AI tools, through trusted resources, and understanding some of the common challenges HCPs have when communicating to their patients.

Learn more

How Datapharm supports AI-generated medicines information

Datapharm distributes medicines information to a variety of key stakeholders in the healthcare ecosystem, including popular AI tools used by UK healthcare professionals. To learn more about how Pharma’s medicines information is delivered throughout a healthcare ecosystem increasingly reliant on AI, contact our team.

Datapharm provides technology-enabled solutions to the global life sciences and wider healthcare sectors. Its mission is to improve the accessibility and effectiveness of medicine safety information – it is the provider of the electronic medicines compendium (emc), which started as a book in the 1970’s and continues to be well embedded in the healthcare setting today.​ 

emc is the resource most trusted by medical professionals to provide accessible, up-to-date, regulated medicine information, including the Summary of Product Characteristics (SmPC), Patient Information Leaflet (PIL), and Risk Minimisation Materials (RMMs). Find out more about emc here.

Over 70 Life Science companies in the UK use their applications for maintaining high compliance standards: emc pi portal, which is helping companies save tens of thousands when updating and sharing their prescribing information, and emc icomply for Field Sales teams sharing product information with HCPs.

If your company is on emc, you may be able to benefit from pi portal or icomply – speak to your company’s Account Manager to learn more: Speak to my Datapharm Account Manager