AI as the NHS Front Door
A Sitora report on safety, accountability and public trust in AI triage across NHS primary care.
AI can help the NHS improve access and reduce pressure. But if AI becomes the first point of contact for patients, it must be independently audited, clinically accountable, transparent and supported by clear human escalation routes.
Published by Sitora. Open for expert comment.
Core recommendation
The NHS should adopt a mandatory AI Trust Test before AI triage becomes a national front door to care.
6
Trust tests
8
Policy asks
30d
Open review
01
The issue
The NHS is moving towards AI-supported triage through digital channels including the NHS App. This could help patients reach the right service faster, but it also introduces new clinical and accountability risks.
02
The risk
If AI incorrectly directs a patient away from urgent care, GP review or emergency treatment, the consequences could be serious. Public trust depends on safeguards before national scale.
03
Sitora’s position
Sitora supports responsible AI adoption in healthcare. Our call is not to stop innovation, but to ensure AI triage is safe, explainable, accountable and trusted by patients.
Sitora position
“The NHS does not just need AI that works. It needs AI patients can trust.”
The report proposes a practical trust framework for AI triage: clinical safety, independent assurance, transparency, human escalation and accountability.
Sitora Framework
The NHS AI Trust Test
Sitora recommends a mandatory trust framework before AI triage is scaled nationally.
Clinical safety validation
AI triage tools should be tested against real-world clinical scenarios, urgent symptoms, complex patients and high-risk presentations.
Independent AI audit
The NHS should not rely only on supplier self-certification. AI systems should be independently audited for safety, bias, accuracy and risk.
Human escalation
Patients must have a clear route to human review where symptoms are serious, unclear, worsening, disputed or potentially urgent.
Clear accountability
Responsibility must be defined before rollout, including supplier, NHS, commissioner, clinical and procurement accountability.
Patient transparency
Patients should be told when AI is being used, what role it plays, what it cannot do, and how they can challenge or escalate a recommendation.
Ongoing monitoring
AI safety cannot be approved once and forgotten. Systems must be monitored for harm, bias, complaints, incidents and model drift.
Policy recommendations
What Sitora is asking for
The report sets out practical safeguards for Government, NHS England, regulators and suppliers.
These recommendations are designed to support responsible AI adoption — not delay useful innovation.
Mandatory AI impact assessments before NHS deployment.
Independent pre-deployment audits for AI triage systems.
Published plain-English safety cases for public trust.
A guaranteed route to human review for patients.
National reporting for AI-related patient safety incidents.
Clear supplier accountability in NHS procurement contracts.
Bias, equality and accessibility testing before rollout.
Clear patient notification when AI influences care routing.
Open expert review
Help strengthen the report before wider submission
Sitora welcomes comments from clinicians, NHS staff, AI researchers, patient-safety specialists, data protection professionals, legal experts, policymakers and patient groups.
Review areas
Evidence base
References and sources
The full report draws on existing policy, regulatory and health AI guidance, including NHS England digital transformation material, MHRA guidance, NICE evidence standards, ICO AI guidance and WHO ethics guidance.
Key reference areas
NHS England — digital transformation, NHS App strategy and digital clinical safety assurance.
MHRA — software and artificial intelligence as a medical device.
NICE — evidence standards framework for digital health technologies.
ICO — AI and data protection guidance.
WHO — ethics and governance of artificial intelligence for health.
Sitora
AI adoption is moving fast. Governance has to move faster.
Download the report, share the policy brief, or contact Sitora to discuss AI governance, assurance and trust frameworks.