Clinical AI Governance Without National Standards
UK clinical AI governance is operating in a documented vacuum. The MHRA National Commission — tasked with producing a unified national framework — has no confirmed publication date for its recommendations, with Q3/Q4 2026 the earliest plausible window. Clinical AI adoption, particularly documentation AI, has already outpaced that timeline.
In the absence of national standards, two reference points now define the operative framework. The GMC has confirmed that existing professional standards apply to AI use — clinicians remain responsible for decisions informed by AI, and deployments must preserve the conditions for those standards to be met: judgement, consent, safety reporting, and escalation. NHS England has published updated governance guidance for ambient scribing (v2, April 2026) — the only use-case-specific NHS guidance currently available — which sets out a concrete procurement and deployment checklist and explicitly acknowledges that liability in NHS AI settings remains complex and uncharted, with Trust exposure a live risk where supplier liability cannot be established.
Tort law analysis confirms primary care as structurally most exposed: GP practices lack the procurement and governance infrastructure to self-assess AI risk, making ICB and federation-level intervention a necessity rather than an option.
The recommended action is "Triage": an immediate internal assessment of all AI tools in deployment or procurement against the GMC standards checklist and the NHS England ambient scribing governance framework. Organisations that wait for MHRA recommendations are accumulating unreviewed exposure month by month. Any governance framework implemented now should have a built-in refresh cycle for when national standards arrive.
£49
