Why have one
AI tools are arriving faster than practices can assess them, some through software updates and others on a locum’s or new staff member’s phone. A policy means everyone works to the same rules, and it’s one of the first things CQC inspectors may ask to see.1
A good policy answers three questions.
Which AI tools are we allowed to use, and for what?
What must never go into an AI tool?
How does a new tool get approved?
What counts as AI
AI is built into everyday software as well as scribes and chatbots. Your policy should cover the whole range.
- AI scribesListen to the consultation and draft notes
- AI triage and online consultationSort requests or suggest urgency
- AI receptionistsAnswer calls or chats, book and signpost
- Results and document toolsFile results, summarise and code letters
- Recall toolsSearch the list to find who to invite
- Chatbots and CopilotChatGPT, Gemini, Copilot in email and Teams
- Call and meeting transcriptionIn Teams or your phone system
- Features added in updatesAI switched on in software you already use
The AI you can’t see
The biggest risk is often the tool nobody told you about: a free chatbot, or a feature that switched itself on.
At Oakfield Surgery: a locum used their own AI scribe app in clinic. Nobody knew where the recordings went. The policy now says only registered tools may be used, on page one of the locum pack.
How it fits with the rest
The policy sets the rules, the register lists what’s approved or live and what’s being assessed, and each tool has its own clinical safety and data protection assessment underneath. NHS England expects a DPIA before AI is used with patient information.2
References
- GP mythbuster 109: Artificial intelligence in GP services. Care Quality Commission
- Information governance guidance: Artificial intelligence. NHS England
Sources on this page last checked 6 November 2026.