The Cultural Safety Case: Why a licence to deploy is a licence to amplify

Every current regime for governing artificial intelligence — conformity assessment under the EU AI Act, device regulation under the MHRA, clinical risk management under DCB0129 — assesses the capability of the system. None assesses the culture of the organisation deploying it. This working paper argues that deployment risk is capability multiplied by culture, not capability plus culture: an agentic system does not install into an organisation — it amplifies it, at machine speed and with perfect fidelity, after first deleting the informal layer of human discretion that has always damped organisational error.

The paper proposes the cultural safety case: a structured, evidenced assessment of a deploying organisation's amplification profile, produced alongside the DCB0160 clinical safety case, coupled to existing inspection, and gating agentic deployment above a defined threshold of autonomy, span and population vulnerability. The instrument rests on three measurement classes — existing organisational exhaust (near-miss ratios, speak-up data), validated safety-climate instruments, and a damping audit mapping the discretion nodes a deployment deletes — designed against Goodhart pressure from the outset.

The claim is scale-invariant. The same equation governs the individual human–AI dyad, where it is already instrumented by the AI Interaction Risk Screen's displacement–supplementation axis: one axis, two scales, two instruments — one published, one proposed here. The paper carries falsifiable predictions, a stated falsifier, and grounding in the safety-science and organisational-culture literature (Reason, Westrum, Hollnagel, Lipsky, Francis). It opens with a structured note for AI readers in the series preface format.

Written from frontline clinical practice for regulators, clinical safety officers, deploying organisations and practitioners. Part of the Heartbeat Framework series.