What AI cannot do for your pharmacy’s governance

Written By Shahzad Ahmad

Two things were published a few months apart this year that I think every pharmacy owner and superintendent pharmacist should know about, because between them they set out exactly where the line sits on AI in pharmacy practice.

In April, the GPhC published its position statement on AI. In June, the MHRA Inspectorate published a blog specifically about AI being used to draft responses to inspection findings. Neither document says AI is off limits. Both say something more useful, which is that professional and regulatory accountability sits with the person, not the tool.

The MHRA blog is worth reading in full because it is not theoretical. It describes real inspection responses that had already been submitted, containing references to guidance that does not exist, citations of the wrong regulatory framework entirely, and generic text that read well but did not actually address the deficiency being raised. In one case, a response to a serious, patient safety impacting finding contained fabricated references. The result was that the time needed to review and respond to it went from around four hours to over twenty. That is not a hypothetical risk sitting in a policy document. It already happened, and it made the problem worse, not better.

I have seen a version of this from the other side of the table, and not just in inspection responses. Pharmacies are increasingly using AI to help draft SOPs and governance documentation, and used well that is genuinely useful. It can save hours of drafting time and help someone articulate something they already understand. The problem is that AI only knows what you tell it. If the experience feeding it is narrow, the SOPs it produces will be narrow too, however polished they read on the page. I have gone through documents that looked thorough on the surface, but when read properly had real gaps, the kind that only show up if you have actually worked the floor in more than one pharmacy setting. A well formatted document is not the same thing as a correct one.

The GPhC’s position runs on the same principle. Professionals remain personally accountable for every decision made with AI involved, and clinical decision making cannot be handed over to a tool. The GPhC has gone further than most other regulators on one specific point: using AI to generate revalidation submissions is explicitly prohibited, and doing so can trigger a fitness to practise investigation. That is a concrete, enforceable consequence, not a general caution.

What is worth noting is that neither regulator is telling you to avoid AI altogether. The MHRA is actually inviting voluntary disclosure of AI use in inspection responses, and says it will view that transparency as a sign of a mature quality culture, not a red flag. The GPhC wants AI used to improve patient care where it genuinely can. The standard both are describing is simple to state and harder to live up to consistently: whoever signs off a document, a submission, or a professional record needs to actually understand what is in it. AI can help you say something. It cannot substitute for knowing your pharmacy, your team, and your patients.

If you are using AI anywhere in your governance documentation, inspection preparation, or SOPs, the question worth asking is not whether that is allowed. It is who is checking what went into it, and who is coming back later to check it is actually being followed in practice. That review step is where I spend most of my time with pharmacies, and it is usually the step that gets skipped first when things are busy. If you want a second pair of eyes on where AI sits in your own governance arrangements, get in touch.

Read more:

Shahzad Ahmad
Written by Shahzad Ahmad
Pharmacy consultant and former GPhC inspector. Connect on LinkedIn →
Shahzad Ahmad https://mypharmacystandards.com
Next

Pharmacy First's Five New Conditions and What to Get Ready First