Practice position
AI assists. People decide.
Use approved systems, minimise information, check important output and keep accountable professional judgement human-led.
Separate the pattern from the person
If you want help improving a case-note template, ask for a blank structure. If you want discussion questions, describe the general topic. If you want a training scenario, build a fully fictional one. These approaches can produce useful material without turning somebody’s life into prompt content.
Removing a name is not enough when age, location, family circumstances, dates or a rare event can identify someone in combination. Proper anonymisation is a technical and contextual judgement, not a find-and-replace exercise.
Use source-led workflows
For guidance summaries, supply an approved public document rather than describing a client. For service communication, provide the approved pathway text rather than referral details. For supervision preparation, ask for generic reflective questions and bring the real practice discussion into established supervision.
Approved systems still need purpose
A managed organisational platform may have different contracts and controls from a personal account, but approval is not a blank cheque. The organisation still needs to define the task, information, lawful basis where applicable, access, retention, review and accountability.
Practical checklist
- Can the task be completed with public information?
- Can a blank template replace a real record?
- Can a fully fictional scenario replace a real case?
- Could details still identify someone in combination?
- Is the platform and precise use formally approved?
Authoritative starting sources
NHS England: AI and information governance ↗ICO: AI and data protection ↗Written by Tony D’Agostino / TD Consultancy. Reviewed . This page provides general professional education, not legal or clinical advice.