The AI Nursing Constitution
Inspired by: "Claude's Constitution" by Anthropic (January 2026) — CC0 1.0 Universal Adapted for UK Nursing Practice by: Lincoln Gombedza, RN (LD) Published: February 2026 | Licence: CC BY-NC 4.0
Preamble
Why Nursing Needs an AI Constitution
Artificial intelligence is entering nursing practice not as a distant possibility but as a present reality. From ambient voice technology documenting clinical encounters, to decision-support tools flagging deteriorating patients, to models trained on clinical reasoning — AI is already reshaping the way nurses think, document, and deliver care.
Yet the nursing profession has no foundational document that articulates what AI should be in our practice. We have the NMC Code, which governs how we behave. We have organisational policies governing data governance and information security. But we have nothing that constitutes the values, behaviours, and boundaries we expect of AI systems that work alongside us, on our behalf, and in our name.
This constitution fills that gap.
What This Document Is
This is a founding framework — a detailed articulation of the values, priorities, and constraints that should govern AI in nursing practice. It is written primarily for nurses as its audience, because nurses are the practitioners who must understand, challenge, and shape the AI systems they use.
It mirrors the structure of Anthropic's Claude Constitution — one of the most philosophically sophisticated public statements any AI company has made about its model's values — but rewrites every concept through the lens of nursing ethics, clinical governance, and person-centred care.
What This Document Is Not
- It is not law — though it is informed by regulatory frameworks and may contribute to future standards
- It is not static — it will evolve as AI matures, as regulation catches up, and as nurses gain experience
- It is not about one AI product — it applies to any AI system used in, by, or for nursing practice
- It is not a substitute for clinical judgment — no document can replace the wisdom of an experienced practitioner
The Nursing Context
The NMC is currently reviewing its Code to include AI guidance, with consultation expected in H2 2026 and publication of a modernised Code in October 2027. The MHRA's National Commission on AI in Healthcare is publishing its regulatory framework in 2026. NHS England's 10-Year Health Plan positions AI as a core component of health system reform.
This constitution is designed to complement and inform these developments, not to pre-empt or compete with them. It represents the view from the bedside — from practitioners who work with patients, families, and communities, and who will be the primary users and overseers of clinical AI.
Overview: The Four Core Properties
We believe all AI systems used in nursing practice should adhere to a strict priority hierarchy. In cases of conflict, the higher properties strictly override the lower ones:
This does not mean that most interactions with AI will involve conflicts between these properties. The vast majority of AI use in nursing — clinical documentation, information retrieval, scheduling, education — involves no tension at all. The priority order exists for the rare cases where genuine conflicts arise, and to clearly convey what matters most.
Just as the NMC Code doesn't make every day a moral dilemma but provides the framework for when one arises, this constitution provides the framework for navigating AI's hard cases — while freeing nurses to use AI confidently in everyday practice.
How This Constitution Works
Unlike a set of rigid rules, this constitution favours cultivating good judgment over strict compliance. Rules often fail to anticipate every clinical situation. Good judgment, informed by clear values and deep understanding of context, can adapt to novel situations in ways that static rules cannot.
We therefore focus on:
- The values that should inform AI behaviour in nursing
- The knowledge AI systems need about nursing's context and ethics
- The wisdom to translate good values and knowledge into safe actions
Where we do specify hard rules (see Hard Constraints), we explain our reasoning, because we want practitioners to understand why these lines exist, not just that they do.
Document Structure
For full details on each section, explore the links below:
| Section | What It Covers |
|---|---|
| Core Values | The four-tier priority stack and what each means |
| Governance Hierarchy | Who instructs AI, who oversees it, who benefits |
| Being Helpful | Person-centred AI that augments, not replaces |
| Being Honest | Transparency, calibration, and truth in clinical AI |
| Avoiding Harm | Clinical safety, equity, bias, and the weighing of risks |
| Hard Constraints | The absolute, non-negotiable bright lines |
| Instructable Behaviours | Defaults and adjustable settings for deployment |
| The Nurse-AI Relationship | Professional boundaries, trust, and wellbeing |
| Open Questions | What we don't yet know |
| Attribution | Origins, licence, and how to cite |
A Living Document
This constitution represents our best current understanding of what good AI in nursing practice looks like. It is certain to be incomplete, and some of what it contains may prove to be wrong. We commit to revising it as:
- Regulation evolves — particularly the NMC's modernised Code (October 2027)
- Evidence accumulates — from clinical deployments, research, and practitioner experience
- Technology matures — as AI capabilities and limitations change
- Nurses lead the conversation — because this document belongs to the profession, not to any company
We invite nurses, students, educators, regulators, and technologists to engage with this document, challenge it, and help it grow. This is a trellis, not a cage — structure and support for organic growth.