FCA authorised · FRN 724952 0117 325 0027Quote & buy →
Aipedia: Veterinary surgeons and practices
The RCVS advice says vets and vet nurses may use AI, but must not wholly delegate professional or clinical decisions to it and must critically appraise every output. It shapes how the RCVS would judge an AI-related complaint and how a PI claim would be defended.
Part of: AI in veterinary practice
In short
The RCVS Standards Committee agreed "Using artificial intelligence (AI) in practice: advice for the profession" in April 2026. It is advisory and creates no new rule; it applies the existing Code. Its core points: decisions must not be wholly delegated to AI; users must critically appraise output and understand the subject; AI scribe records must be manually verified (Chapter 13); developer access to client data and training on live data must be checked (Chapter 14); AI tasks go only to trained staff; and practices should run an adoption checklist. It says nothing about liability or insurance.
The AI advice was agreed in April 2026 and published in the Spring 2026 standards update, alongside a revised Chapter 13 on clinical and client records. The RCVS explains that the Code and supporting guidance "do not specifically refer to the use of AI", so the advice shows how existing principles apply. The Standards Committee deliberately avoided "additional, onerous obligations".
That makes the advice guidance rather than a new standard. In practice it still carries weight: if a concern about AI use reaches the RCVS, the Code duties are what will be tested, and the advice shows how the RCVS reads them.
| Item | Date | Status |
|---|---|---|
| AI advice for the profession | Agreed April 2026 | Advisory, applies the Code |
| Revised Chapter 13, records | Approved April 2026 | Supporting guidance |
| AI Roundtable report | Published August 2024 | Discussion report, no timetable |
| Joint statement on AI in education | February 2026 | Joint statement with five other regulators |
The central line is that "professional and clinical decision making must not be wholly delegated to AI". A human stays in the loop and stays responsible. This echoes the 2024 roundtable, where over 100 delegates agreed vets should keep ultimate responsibility. AI can draft, suggest, measure or flag, but the decision to diagnose, treat, prescribe or certify remains the vet's.
The roundtable report floated further options, including Code amendments, accreditation standards, day-one competences and training, but set no timetable, and none had been adopted when this page was written. In February 2026 the RCVS also signed a joint statement with five other regulators on AI in education, which keeps learners responsible for their own work.
Vets and vet nurses are responsible for choosing to use a tool, and the RCVS asks them to appraise AI output the way they would appraise research: where it came from, what data it was built on, its assumptions and its limitations. The user needs "sufficient understanding of the subject matter" to judge whether the output is right.
The advice names three risks: bias, hallucination, and "overreliance or overconfidence" that can erode clinical skills. This links to Code 1.2, keeping within your own area of competence, and 3.3, maintaining knowledge and skills. A tool does not extend what a vet is competent to do.
Output from an AI scribe "should be manually verified" and errors corrected promptly. Where AI helps with a differential diagnosis, the record should hold the clinically relevant information, not the full list the tool generated. Code 2.5 still requires clear, accurate and detailed records.
The revised Chapter 13 also requires vets and nurses to be identified in records (13.9) and records to be kept "for as long as is necessary, taking into account legal and regulatory responsibilities" (13.10). The Standards Committee papers noted PI policy terms as one reason records may need to be kept.
Code 2.6 prohibits disclosing information about a client or their animals to third parties. The advice asks practices to check that developers cannot access practice data, or train on it, without consent; to check whether a tool learns from live client data and, if it does, that clients consented; and to be "open and honest" with clients about how AI processes their data. It points to ICO guidance for the data protection side.
AI tasks should be delegated only to staff with "appropriate knowledge and training", and lay staff should be reminded of their confidentiality obligations. Before adopting a tool, the RCVS suggests checking:
For most practices the advice formalises what good clinical governance already requires: know what tools are running, check outputs, keep honest records and protect client data. The practical gap is evidence. If a complaint or claim arises, the vet will want to show that the output was reviewed by someone competent and that the tool was assessed before use.
The advice is silent on liability and insurance. Code 3.4 still requires PI or equivalent arrangements for all professional activities, and AI-assisted work is a professional activity. Whether a particular policy responds without dispute depends on its terms. See does veterinary PI cover AI mistakes and the RCVS PI requirement. Related guides: AI scribes and records, client data and AI and delegating AI tasks.
Apex arranges PI for vets and practices. A broker can review whether your cover reflects the AI tools you have adopted and how AI-assisted work is treated in the wording. More on AI in veterinary practice.
It is advice, not a new rule. It explains how the existing Code and supporting guidance apply to AI. Because the Code is what the RCVS applies to concerns about conduct, following the advice is the sensible way to show those duties were met.
Not for every use. The advice asks practices to be open and honest about AI data processing and to obtain consent where a tool learns from live client data. Developers should not access or train on practice data without consent.
The RCVS says record the clinically relevant information rather than the generated list. Note what you considered and why, in your own clinical terms, so the record shows your reasoning rather than the tool's output.
Yes. It is addressed to vets and vet nurses, who are both responsible for choosing to use a tool and for appraising its output. Delegated AI tasks should go only to staff with appropriate knowledge and training.
Tell us about your firm and the work you do. A broker will review how your current PI treats AI-assisted work, or arrange new cover. No fee for a quote. Or call 0117 325 0027.
Start a PI quote Request a callbackApex Insurance Brokers Limited is authorised and regulated by the Financial Conduct Authority. Registered in England and Wales, company number 07014570. This page is general information about how AI is used in the profession and how insurance may respond. It is not legal, regulatory or insurance advice on your circumstances, and it does not guarantee that cover will be available or on what terms.
