AI and the CQC: what inspectors expect, and how to get it right

Visiting the Care Show London this year it was clear that AI is a big focus within the sector. Across the conference programme 11 of the scheduled talks has AI within the title, and it was evident AI innovation companies had a very visible presence. Did you spot the ‘Ask how AI can help your service’ t-shirts!

AI is already part of adult social care in England, and CQC inspectors are starting to write about it. At Care Surveys we reviewed 3,322 CQC assessment reports for care homes, nursing homes and home care services published in 2026. Of those, 69 mention AI or AI-driven technology.

The tools inspectors described range from apps that spot pain in people living with dementia to systems that scan care records for emerging risks, and sound sensors that alert night staff when someone needs help. None of the 69 reports criticised a provider simply for using AI. What inspectors did look at was how it was used, who checked it, and whether people had agreed to it.

This guide sets out where the CQC stands on AI, how existing regulations already apply to it, and what inspectors are likely to look for. It ends with a practical checklist you can start on this week.

1. Why this matters now

AI has moved beyond back-office admin and into care itself. The CQC reports we reviewed describe AI being used to capture initial assessments, draft care plans, check medication records, analyse incidents and flag changes in people's needs. Those are the tasks the regulations care about most.

The risks are real too. In February 2026, the Ada Lovelace Institute reported that AI transcription tools used by social workers in 17 councils had produced errors in official records.

Earlier research from the University of Oxford found some care providers using free chatbots such as ChatGPT to write care plans. That puts personal data at risk and can lead staff to act on inaccurate or biased information.

The message for providers is simple; AI can free up time for care, but the provider stays accountable for every decision it informs.

2. The CQC's current position

On 21 May 2026, the CQC published its first formal statement on AI: Artificial intelligence in health and social care: CQC's role, expectations and plans. Its stance is supportive, but conditional.

• It encourages AI use where the technology benefits people and results in more effective and efficient services.
• It does not approve AI products: the CQC regulates providers, not technology. Choosing a tool, and using it safely, is your responsibility.
• AI does not decide your rating; the CQC says the absence or presence of AI does not predict a specific rating.

Our review backs this up; a quarter of the services whose reports mentioned AI were rated Outstanding, well above the sector average, but seven services rated Requires improvement also had their AI use noted. Inspectors credit AI when it is part of a well-run service; it does not make up for weaknesses elsewhere.

3. How existing regulation covers AI

There is no separate AI rulebook for adult social care. Instead, the regulations you already work to apply to any AI you use. The CQC names three as particularly relevant:

The CQC also sets out principles for good use of AI:

• Support, don't replace: AI can enhance human decision-making but not replace it.
• Human oversight. AI outputs and processes are continuously monitored and evaluated.
• Transparency and choice. People understand how AI is used in their care, and non-digital routes are offered where needed.
• Safety, reliability and security. Tools are safe, accurate and protect people's data.

What's coming

The CQC's 2026/27 priorities include understanding how AI affects care, deciding what further guidance is needed, planning how to respond if AI causes risks, and sharing best practice. It has also said it may develop tailored AI guidance for different sectors and settings.

The direction of travel is clear; in its draft framework for GP practices, the CQC gives staff implementing AI in clinical care pathways as an example of outstanding practice. Social care providers should expect similar thinking in their own framework.

4. What inspectors are likely to look for

Inspectors want to see that AI is governed, not just installed. Be prepared to answer these four questions:

1. Where is AI used? Can you list every tool, and what it does?
2. Who checks its outputs? Is there a named person who reviews anything that informs a care decision?
3. Were people told, and did they agree? Is consent recorded, and is there a non-digital option?
4. What happens when it gets something wrong? Are errors spotted, logged and learned from?

In our review, 56 of the 69 reports that mentioned AI did so under Well-led. That makes sense: how you govern AI is a leadership question. Inspectors also mentioned AI under Safe and Effective, usually where it helped manage risk or plan care.

What good looks like in recent reports

• Consent and training - At a care home rated Good in March 2026, people's permission for night-time AI monitoring was recorded in their care plans. Staff also had specialist training on the system and how to respond to it.
• Human checks on AI outputs – Leaders at a service rated Outstanding tested AI for audits and meeting records, and recognised the results still needed checking for accuracy.
• AI alongside professional judgement - At another Outstanding service, managers used an AI tool alongside professional judgement to review care records after people were discharged from hospital.
• Consent paperwork - At one nursing home rated Good in April 2026, inspectors found mental capacity assessments for acoustic monitoring and bed sensors were incomplete or missing. Whilst the technology was helping people; the paperwork behind it was not in order.
• Technology without the fix - A home care service rated Requires improvement in September 2026 had brought in an AI rota system to tackle late calls. Inspectors noted it was a positive step but not yet fully effective.

Where it fell short

Inspectors are using AI too; the CQC's chief inspector of adult social care has said it will be a productivity tool that frees inspectors to spend more of their time observing care. So, what they see on the floor will matter even more.

5. What providers should do now

Don’t wait for formal guidance, these eight steps cover what the CQC has already said it expects:

1. Keep an AI register - List every tool in use, including staff using free AI platforms informally (including those built into operating systems such as CoPilot and Gemini).
2. Write an AI use policy and complete a DPIA for each tool - A data protection impact assessment shows you have thought through the risks to people's information before you start.
3. Tell people and get consent - Prepare plain-English information for people and families and keep a non-digital option. This includes the use of AI outside of care delivery, such as if AI notetaking is used within interview processes for example.
4. Require human sign-off - Anything that informs a care decision should be checked by a named person. AI has the ability to flag a risk, but staff should be deciding what to do about it.
5. Train staff on the specific risks - Generic "digital skills" training isn't enough anymore. Staff need to know that AI can make things up, miss context or reflect bias, and what to do when it does.
6. Monitor outputs and log AI errors as incidents - Treat a wrong AI output like any other near miss: record it, review it and share the learning.
7. Never let AI write your policies unedited - Compliance specialists warn that generic, AI-drafted policies are a red flag for inspectors and can undermine confidence in your leadership. Your policies must reflect how your service really works.
8. Do due diligence on suppliers - Ask where data is stored, how the tool was tested, and how errors are reported.

A final note

AI feels unavoidable; it’s on our phones, built into operating systems and programs we have used for years, and new technology services seems to pop up daily offering to solve all our problems.

It certainly has its place within care; where it can save providers time, not to replace staff, but giving them more time to provide amazing care.

Innovation is exciting, and there are some great success stories, which as the technology evolves and improves, we would expect to see more of. But it will always carry risks. Using the checklist above can help minimise risk and give you confidence when introducing new systems and software. This preparation will also help you get ahead before the CQC, quite possibly, implements AI specific regulations in the future.

How we reviewed CQC reports - We used the CQC's published care directory to identify every care home, nursing home and home care service with a CQC inspection dated in 2026, as of 7 October 2026. We then searched the latest assessment report for each, 3,322 reports in total, for AI-related terms and reviewed each mention by hand. Older PDF-only reports were not included.