# England issues non-statutory AI guidance for social care

Published 2026-11-20 · AI.info Team · AI.info

> New guidance says AI should not make decisions about risk, eligibility or interventions in children's and adult social care. Practitioners should review AI-generated drafts and use approved tools under their organisations' policies.

AI should not make decisions about risk, eligibility or interventions in children’s and adult social care, according to new government guidance for England. The Department for Education and the Department of Health and Social Care published the guidance on 20 November, alongside a quick guide for practitioners.

Source [gov.uk](https://www.gov.uk/government/publications/artificial-intelligence-ai-in-social-care)

The guidance is non-statutory and leaves decisions on whether and how to adopt AI to individual organisations. It says the advice should be applied alongside local policies, professional standards and existing legal and regulatory requirements.

## Where AI may assist

It is aimed primarily at local authorities providing children’s and adult social care, although its principles also apply to other organisations offering social care and related services. The guidance addresses frontline practitioners, supervisors and managers as well as staff working in areas such as data protection, procurement and digital transformation.

The departments say AI may help with speech-to-text transcription of visits and meetings, or with drafts of case notes, reports and letters. Practitioners should review those drafts and add relevant observations, including body language, tone and aspects of the physical environment that a tool might miss.

The practitioner guide suggests other possible uses, including bringing information together before an assessment, identifying gaps in records, simplifying complex information and translating text. It says AI may help organise information or prepare a draft assessment or care plan, but the analysis, conclusions and recommendations should reflect the practitioner’s own judgement.

## Decisions stay with practitioners

Practitioners should treat AI output as a first draft, checking and amending it before use, the guidance says. It identifies inaccuracies, bias and undue reliance on AI outputs among the risks that organisations need to manage.

The quick guide says checking AI-assisted work involves more than proofreading: practitioners should test whether it is accurate, supported by evidence and reflects their judgement. Where appropriate, they should consider letting the person receiving care, or their family, see a transcript or summary so their voice and needs are captured accurately.

## Data and oversight

The guidance says organisations should have, or be working towards, governance arrangements covering acceptable use, data handling, training, supervision and review. They should make clear which tools are approved and the conditions for using them.

Personal information should not be entered into publicly accessible AI tools, and staff should follow their organisation’s data protection policies. Organisations should also provide accessible policies explaining how AI use is disclosed, what choices people have and how objections are handled.

The guidance also warns that poorly implemented AI may reduce opportunities for meaningful engagement and undermine trust. It asks organisations to consider people who may have limited digital access or need help engaging with technology, and to consider involving people receiving care in service design or governance.

The departments expect further resources and support to build on the guidance in future.

## Sources

1. Original source · Government or agency page [GOV.UK · Artificial intelligence AI in social care](https://www.gov.uk/government/publications/artificial-intelligence-ai-in-social-care) gov.uk

Other pages read

2. [GOV.UK · AI in social care](https://www.gov.uk/government/publications/artificial-intelligence-ai-in-social-care/ai-in-social-care) gov.uk
3. [GOV.UK · Using AI in social care a quick guide for practitioners](https://www.gov.uk/government/publications/artificial-intelligence-ai-in-social-care/using-ai-in-social-care-a-quick-guide-for-practitioners) gov.uk

## Related stories

- [EU asks AI deployers to test a draft rights-impact questionnaire under the AI Act](https://ai.info/news/eu-feedback-ai-rights-impact-assessment-template)
- [CHIME says 93.2% of surveyed health organizations use AI](https://ai.info/news/chime-2026-health-systems-ai-use-governance)
- [England’s university regulator calls for more AI-free assessments](https://ai.info/news/england-university-regulator-ai-free-assessments)
- [Nadella says a lot of critical infrastructure cannot be patched](https://ai.info/news/nadella-unpatchable-critical-infrastructure-ai)
- [Ofqual seeks wearable-tech training and plans A-level AI proposals](https://ai.info/news/ofqual-wearable-technology-ai-assessment-warning)

## Explore

[More articles](https://ai.info/news)

---

Canonical: https://ai.info/news/england-ai-guidance-social-care
