AI Trust & Data Protection

Governed AI for children’s social care.

IndiCare Intelligence trust framework

IndiCare Intelligence is designed to support adults, protect sensitive information and keep professional judgement at the centre of care.

ORB Residential supports professional judgement. It does not replace safeguarding procedures, statutory responsibilities, provider policies or human decision-making.
Trust by design

Built for regulated care, not generic AI use.

The AI trust approach is built around professional boundaries, provider control, data minimisation, redaction, safe audit and human review.

Human accountable

Adults remain responsible for decisions, records, escalation and actions.

Provider controlled

Providers should control external AI, storage, redaction and usage settings.

Child-centred

The aim is to support safer care, clearer records and better professional thinking.

Our principle

AI supports professional judgement. It does not replace it.

ORB Residential is built to support adults working in children’s homes. It can help draft, structure, reflect and identify possible gaps — but the adult remains accountable.

Help adults see the child more clearly, record more carefully and act more confidently — without removing human responsibility.

Provider control

Providers control how AI is used.

IndiCare is designed so providers can decide how external AI, redaction, storage, voice, dictation and usage audit are managed.

External AI

Providers can control whether external AI processing is enabled. External AI should not be assumed as the default in regulated care settings.

Redaction

Redaction helps reduce unnecessary sharing of names, contact details, dates of birth and other sensitive identifiers before external processing.

Prompt storage

Prompt storage should be off by default unless a provider explicitly enables it with appropriate governance, policy and user understanding.

Transcript storage

Voice or dictate transcripts should not be stored by default unless enabled and governed by the provider. Spoken content can be highly sensitive and should be treated carefully.

Usage audit

Providers can review safe usage metadata such as feature, model, redaction status, storage status and general usage patterns without exposing raw sensitive content unnecessarily.

Approved use

Providers should decide what ORB is approved for, what requires manager review, and what must always be handled through existing safeguarding or operational procedures.

Boundaries

What AI is not allowed to decide.

IndiCare is designed to support thinking and recording, not make statutory, clinical, legal or professional decisions.

No safeguarding decisions

ORB can support reflection, but it does not decide safeguarding thresholds or replace local safeguarding procedures.

No diagnosis

ORB does not diagnose children, assess mental health conditions or replace health professionals.

No Ofsted grade prediction

ORB can support inspection readiness, but it does not predict Ofsted grades, inspection findings or outcomes.

No LADO decision

ORB can remind adults to consider the right route, but it does not decide whether a LADO referral is required.

No police decision

ORB does not decide whether police action is required. Adults must follow procedure and professional advice.

No legal decision

ORB does not provide legal decisions, formal legal advice or replace legal professionals.

Data protection

Built around minimisation, redaction and review.

Children’s homes hold highly sensitive information. IndiCare is designed to use only what is needed and keep humans in control.

IndiCare is designed to protect

  • child-identifiable information
  • staff information
  • family and contact details
  • health and safeguarding information
  • provider records
  • local policy information
  • voice and dictate transcripts
  • records that may form part of a child’s life story

Safeguards may include

  • provider AI controls
  • redaction before external processing
  • safe usage audit metadata
  • prompt and transcript storage controls
  • human review before use
  • trusted source governance
  • approved use policies
  • clear professional boundaries
How safe use works

A simple model for responsible AI use.

ORB should be used as support for thinking and drafting, not as the final decision-maker or final record without adult review.

1

Ask

The adult asks ORB a question, dictates a note or requests help with wording.

2

Support

ORB helps structure thinking, highlight possible gaps and draft safer language.

3

Review

The adult checks accuracy, context, tone, actions and whether procedure has been followed.

4

Decide

The human decides what to do, what to record and what must be escalated.

Trusted sources

Governed guidance, not random scraping.

IndiCare is designed around trusted source governance. Local policies and safeguarding procedures should be uploaded, approved and governed before use.

National guidance

Statutory and regulatory guidance can support the foundation of ORB’s responses when appropriate and kept under review.

Local procedures

Local safeguarding and provider policies should be approved before they are relied on for practice guidance.

Human approval

High-risk source updates should be checked before becoming part of live practice guidance.

Implementation

Trust is not just a feature. It is how ORB should be introduced.

Safe AI in children’s social care depends on training, boundaries, governance and honest communication with staff.

Clear staff guidance

Staff should know what ORB can support, what it cannot decide, what information should not be entered, and when manager or safeguarding escalation is required.

Manager oversight

Managers should review how ORB is being used, whether outputs are being checked properly, and whether staff are relying on it safely.

Provider policy alignment

ORB should sit alongside provider policies, data protection expectations, safeguarding procedures and quality assurance arrangements.

Learning without exposing children

IndiCare should learn patterns, not children. Improvement should focus on anonymised, aggregated and practice-level learning rather than identifying individual children.

FAQ

Questions providers may ask.

These are the questions we expect providers, managers, RIs, commissioners, investors and safeguarding-minded partners to ask about IndiCare Intelligence.

Does IndiCare use OpenAI or external models?

IndiCare may use external model providers where enabled and appropriate. External models may support drafting, summarising or language generation. IndiCare controls should sit around privacy, redaction, sources, audit and human review.

Can providers turn external AI off?

Yes. IndiCare is designed so providers can control whether external AI processing is enabled.

Are prompts or transcripts stored?

Prompt and transcript storage should be off by default. If enabled, this should be a deliberate provider decision with appropriate governance, staff understanding and data protection controls.

Does ORB replace staff or managers?

No. ORB supports adults. It does not replace professional judgement, safeguarding procedures, registered managers, responsible individuals, social workers, LADO, police, health professionals or Ofsted.

Does IndiCare learn from children’s data?

IndiCare should learn patterns, not children. Learning should be anonymised, aggregated and focused on improving practice rather than identifying children.

Can ORB give safeguarding advice?

ORB can support safeguarding thinking, recording and reflection, but it should not be treated as a safeguarding decision-maker. Adults must follow local safeguarding procedures, provider policy and statutory responsibilities.

Can ORB write records for staff?

ORB can help draft and improve wording, but the adult must review, correct and approve anything before it is used. The adult remains responsible for accuracy, context and professional judgement.

Why is AI trust so important here?

Residential children’s homes hold sensitive information about children, families, staff and safeguarding concerns. AI must therefore be governed carefully, introduced safely and used to strengthen professional practice rather than replace it.

Trust, safety and professional judgement.

IndiCare Intelligence is being built to support adults, protect children’s data and keep human accountability at the centre of care.