AI is now producing the accounts that govern people’s lives. The question nobody is asking is whether the meaning survived. DNC provides the governance architecture and workforce readiness to address that gap — before deployment, mid-implementation, and at every point where meaning is at risk.
In healthcare, policing, social care, and legal proceedings, AI systems are already shaping the records that govern people's lives. Consultation transcripts. Police interview summaries. Clinical notes. Social care assessments. These are constitutive records — they don't just describe what happened, they determine what happens next.
The problem is not that AI is used. The problem is that interpretation is being productised before it is governed. Systems summarise, compress, and route narrative without anyone asking: does this summary carry the meaning that was actually present? Who authorised this compression? What is lost?
This is the meaning risk. And it arrives quietly — embedded in workflows, before scrutiny is even possible.
DNC is a governance and workforce readiness framework that addresses this gap at every stage — before deployment, where interpretive accountability needs to be designed in; mid-implementation, where organisations already using AI tools need provenance architecture; and at workforce level, where practitioners need the judgment to interrogate what the system gives them. The intervention point depends on where you are. The concern is the same throughout.
DNC is structured around three pillars of care — delivered through a suite of named instruments deployable at commissioner, builder, or workforce level.
Asks what was written, inferred, versioned, tagged, and silently transformed. Beneath it sits evidential care — the obligation that source materials, provenance, and reasoning remain inspectable.
DNC asks whether the record is faithful to the person — not just internally consistent.
Asks whether the record remains faithful enough to the person, the context, and the judgment process to be safely acted upon by somebody else. Beneath it sits interpretive care — the obligation not to confuse fluency with judgment.
DNC treats distortion of meaning as a governance failure, not a soft issue.
Asks what happens as a record travels — through decisions, panels, appeals, references, and institutional memory. Beneath it sits aftercare — the obligation that live routes for challenge, correction, annotation, and redress exist once the record starts doing work in the world.
DNC asks what the system makes possible later — not only what it does now.
DNC is sector-agnostic by design. The meaning risk is the same wherever AI systems generate records that govern people's lives.
DNC is also a place of connection for people working with these questions in real settings. Not a network for its own sake. Not a forum for hype or posturing. A space for people who are trying to protect dignity, context, and trust in systems shaped by AI and digital tools — and who want language, thinking, and company for that work.
The aim is to build shared practice: a place to develop language, compare approaches, surface what is working and what is not, and think carefully about what humane digital work actually requires on the ground.
The community spans healthcare, policing, social care, education, financial services, and public services. What it shares is not a sector but a concern: that the person remains recognisable in the record, and that someone is responsible when they are not.
Across every sector moving AI into its workflows at scale, the same pattern is emerging. Digital systems do not simply carry information. They shape how people are understood, remembered, categorised, and acted upon. And the governance infrastructure being built around those systems — however rigorous — is not yet asking what happens to the meaning of the story itself.
DNC is part of a wider recognition of that gap. It is not a slogan or a finished doctrine. It is an emerging discipline for people who can see what is missing — and who want something practical to do about it.
That will not happen through one framework or one organisation. It will happen when enough people in enough institutions begin to name the same thing, hold the same standard, and refuse to treat meaning as a soft issue.
Stephen Hall is a writer, consultant, educator, and facilitator whose work brings together culture, care, communication, and institutional practice. Over more than two decades he has worked across higher education, performing arts leadership, simulation-based learning, healthcare education, and wider public-service-facing consultancy. What connects that work is a sustained concern with how people are listened to, how their accounts are interpreted, and what gets lost when complex human situations are turned into records, summaries, and decisions.
He is the founder of Digital Narrative Care (DNC), a framework and emerging practice for protecting meaning as stories travel through digital and AI-mediated systems. Developed through work spanning healthcare, higher education, and public services, DNC asks what happens when systems produce accounts that are fluent, plausible, and operationally useful, but wrong about what matters. Its core concerns are record integrity, meaning integrity, and temporal integrity — and its purpose is simple to state even if harder to do: helping institutions keep meaning human.
Three established series: Taking Care (editorial and analytical), The Assurance Dial (practice and governance), and Making Human (Sunday arts). Welcome to the Machine is an occasional sector series, published when the evidence warrants it.
Subscribe — FreeDNC is built for commissioners who need governance architecture, builders who need workforce readiness frameworks, and policy colleagues who need an upstream argument with evidential grounding. If any of that is your context — the conversation is open.
POLICY BRIEF — APRIL 2026
The upstream governance gap in public-sector AI deployment.
Download PDF →