AI Is Already in Social Care: What the Care England Research Tells Providers About the Governance Gap

Direct answer

Research published by Care England in October 2026, led by Dr Caroline Emmer De Albuquerque Green of the University of Oxford's Institute for Ethics in AI, found that adult social care providers surveyed are already using AI across care planning, medication management, administration and falls prevention technology. The biggest challenge identified was not cost or technology. It was not knowing how CQC will inspect AI use in practice.

At a glance

Providers are already using AI daily across care planning, administration and medication management.
Lack of regulatory clarity was identified as the biggest challenge.
AI can increase administrative workload when outputs require verification.
Providers reported feeling unsupported despite having internal policies.
Care England calls for the Casey Commission to address power relationships across the sector.
DHSC published adult social care AI guidance in the same week.

What the research found

The Care England study, AI has arrived in social care services but more needs to be done to support providers with adoption, examined the experiences of care leaders through a survey of 26 providers and interviews with a further 20. The research was carried out with the University of Oxford's Institute for Ethics in AI, Serene Care, the Digital Care Hub and Care England.

Respondents described current applications including service management, recruitment, drafting care plans, analysing care notes, medication management and AI-enabled sensors to help prevent falls. Most respondents considered their organisations fully digitised, and the majority of those using AI said they did so daily.

AI is therefore not a future prospect for social care. It is a current operational reality that providers are managing without the governance infrastructure they say they need.

The governance gap: what providers actually said

Providers identified lack of regulatory clarity as the biggest challenge to successful AI adoption, rather than cost, training availability or technology reliability. They did not feel sufficiently supported or empowered by regulators and commissioners, and were uncertain about how CQC inspectors would be trained on AI, how inspection decisions would be made, and what would constitute adequate evidence of responsible use.

This is a gap in clarity rather than willingness. Many providers had introduced internal policies and governance arrangements, but lacked practical guidance about what good governance looks like and how it would be assessed.

National activity is developing. CQC has set out its role in ensuring AI contributes to safe, person-centred and equitable care; DHSC published guidance on AI in adult social care; and health and social care regulators have issued a joint statement of intent. Yet providers deploying AI daily report that national principles have not translated into operational clarity at service level.

The workload problem nobody expected

AI can increase rather than reduce administrative workload. Providers described checking AI-generated outputs for accuracy, bias, discrimination and completeness. Managers also reported additional time spent training staff, while some leaders worried that important information could be missed or that over-reliance on generated content could undermine person-centred care.

The administrative burden of verification can become a barrier to adoption. Professionals remain responsible for reviewing and verifying AI-generated outputs before they are used in care. That responsibility creates a permanent professional obligation, not merely a temporary transition cost.

This is why the distinction between AI that generates content and AI that structures what a professional already knows matters in practice. Organising a professional's own observations into a compliant format is different from generating content from a prompt that the professional did not produce.

What the research means for providers right now

For registered managers and care providers currently using AI, five governance actions address the gap between adoption and oversight:

Adopt an AI policy. DHSC guidance recommends that organisations introduce an AI policy, even if that policy prohibits work-related AI use.
Maintain an AI systems register. Record every AI tool in use, approved or otherwise. You cannot govern what you do not know is being used.
Train staff. Establish which tools are approved, which outputs require verification and what to do when content is inaccurate or incomplete.
Document output verification. Create a review record showing that AI-generated content was checked before entering care records.
Complete supplier due diligence. Record what each tool does, how it handles personal data and what its limitations are.

Build the governance infrastructure

ReporticaAI's AI Governance Management Pack includes an AI Policy, AI Systems Register, Risk Assessment, Acceptable Use Standard, Output Review Record, Supplier Due Diligence, Staff Training Log and Review Schedule.

View the AI Governance Management Pack

The provider request the research makes most clearly

Care England's research calls for the resources and support necessary to train staff appropriately. That support remains uneven. Providers are making governance decisions using their own judgement, resources and interpretations of guidance that does not always translate directly into operational practice.

The governance infrastructure does not need to wait for the Casey Commission's recommendations or a future CQC inspection framework. The components are available and implementable now. Their value is not only compliance; they allow a service to demonstrate at inspection that it understood the risks AI creates and managed them deliberately rather than by default.

Sources

Care England / University of Oxford, AI has arrived in social care services but more needs to be done to support providers with adoption, October 2026.
Department of Health and Social Care, Using AI in adult social care, October 2026.
MHRA, MHRA clarifies regulatory status of ambient voice technologies used in the NHS, July 2026.
National Commission into the Regulation of AI in Healthcare, recommendations for a future regulatory framework, September 2026.

This article is published in accordance with PAIDS™ (Professional AI Documentation Standards): well-sourced, researched and defensible with verifiable data.