Who Owns the Risk When an AI Scribe Writes Something Nobody Said?

Who owns the risk when an AI scribe writes something into a patient record that nobody actually said?

The Health Services Safety Investigations Body (HSSIB) has today launched a patient safety investigation into the use of Ambient Voice Technology (AVT) in hospitals.

Focused on acute adult secondary care, it will look at:

• Whether and how AVT has already contributed to patient harm, and where it could
• The specific safety risks the technology introduces
• Where accountability for AVT safety actually sits, locally and nationally

Three Things Stood Out in HSSIB's Intelligence Review

1. Deployment is running ahead of our collective understanding of the safety implications.
2. The national conversation has weighted efficiency gains far more heavily than safety risk.
3. The ways we spot and report AI-related incidents may not yet be robust enough for emerging risks to surface at all.

That last point is the one I would sit with. Where reporting routes are immature, an absence of reported incidents is not evidence of safety.

The Timeline Problem

Then look at the timeline. HSSIB expects to report in summer 2027. Regions are procuring AVT at scale now. And the consultation on DCB0129 and DCB0160 closes on 11 September 2026 - ten months before that report lands. The chance to shape the standards arrives before the evidence does.

So the question for boards deploying AVT this year is not whether to wait for HSSIB. It is whether your hazard log, your named Clinical Safety Officer and your incident reporting route would let you recognise an AVT-related harm if one occurred.

HSSIB is asking to hear from patients, families, carers and professionals before publication - contribute your experience.

TrustPoint Digital Health provides independent clinical safety governance for AI and digital health products under DCB0129 and DCB0160. Get in touch to discuss your safety case.