3 August 2026 · CareTime
If your care home is already using AI — even something as everyday as ChatGPT to draft a policy — the Care Quality Commission now has clear expectations for how you should be doing it. In May 2026, CQC published its position on artificial intelligence in health and social care, and on 29 July 2026 national guidance for the responsible and safe use of AI in health and social care services followed. The message is consistent: CQC encourages AI where it makes care safer and more person-centred, but it expects providers to show training, governance, data protection, and human oversight behind every tool they use.
For a registered manager, that turns AI from a quiet efficiency into something an inspector may ask about. The good news is that the expectations are practical, not technical, and most of them are habits you already apply to other systems in your home. This guide sets out what CQC has said and what it means for the way you choose and evidence AI.
CQC's published position is that it welcomes innovative technology, including AI, where it benefits people and leads to more effective, person-centred care. It does not approve or certify individual products. Instead, it regulates the outcome — whether the care people receive is safe, effective, caring, responsive and well-led — and treats the AI you use as part of how you achieve that.
Alongside CQC's own statement, the July 2026 national guidance for the responsible and safe use of AI reinforces the same principles across health and social care. Taken together, they point to four things CQC expects providers to be able to show.
The people using an AI tool should be trained and confident with it, and you should have considered whether it fits your existing ways of working rather than cutting across them. An AI tool nobody understands is a risk, not an improvement.
CQC expects mechanisms — risk assessments, and a way to recognise, report and investigate when something goes wrong — so that AI contributes to high-quality, equitable care. This is Well-Led territory: who signed off the tool, who reviews it, and what happens if it fails.
Where an AI tool handles personal data, CQC expects a Data Protection Impact Assessment (DPIA) that documents the risks to people's privacy and how you have minimised them. This is a legal requirement under UK data protection law, not a nice-to-have.
This is the one that matters most for care. CQC expects human oversight to be retained so that professional judgement sits behind every decision. AI can inform a decision; it should not make it unaccountably. A person remains responsible.
Adoption has run ahead of governance. Birdie's July 2026 research into UK homecare providers found that around 70% are already using AI in some form, yet only around two-thirds have any formal policy governing how they use it. Much of that use is general-purpose consumer tools — ChatGPT, Microsoft Copilot — rather than software built for care. (This data is drawn from homecare rather than residential settings, so treat it as directional, but the governance gap it describes applies across the sector.)
That gap is exactly what CQC's expectations are pointing at. A home using AI without a policy, a risk assessment, or a DPIA is not breaking a specific rule today — but it would struggle to evidence good governance if an inspector asked. Care England has separately warned that AI-generated policies, produced without proper oversight, can put a CQC rating at risk when they don't reflect what the home actually does.
The distinction CQC draws — AI supporting judgement rather than replacing it — is a useful test when you weigh up any care technology. Tools that quietly make decisions on your behalf carry the most governance weight. Tools that surface information for a person to act on carry the least, because a human stays in the loop by design.
This is where CareTime sits. Silent Guard monitors and logs every incoming call to your home and delivers a daily Morning Brief to your registered manager by 8am. It does not decide who to call back, judge a safeguarding concern, or resolve a complaint. It produces a factual, timestamped record and puts it in front of the person whose job it is to decide. The professional judgement — and the accountability — stays with your team.
That design maps neatly onto CQC's expectations. Human oversight is built in. The output is a factual record rather than a generated document, which sidesteps the Care England warning about AI-written policies. And because the record is dated and consistent, it doubles as the kind of contemporaneous evidence the Well-Led and Responsive key questions ask for.
You do not need a technology strategy to meet CQC's expectations. For each AI tool you use, you should be able to answer four questions: who is trained to use it, who governs it and reviews it when something goes wrong, whether a DPIA covers any personal data it touches, and where the human oversight sits. If you can answer those for every tool — including the consumer ones staff may already be using informally — you are in a strong position.
Does CQC approve or certify AI products for care homes? No. CQC does not approve individual products. It regulates the quality and safety of the care you provide and treats any AI you use as part of how you deliver it. Responsibility for choosing and governing a tool stays with the provider.
Do we need a written AI policy? CQC expects effective governance, and a written policy is the clearest way to evidence it. Birdie's July 2026 research suggests only around two-thirds of providers currently have one, so a clear policy is also a way to stand out under Well-Led.
Does a call monitoring tool like Silent Guard need a DPIA? Any tool that processes personal data should be covered by a Data Protection Impact Assessment. CareTime can provide the information you need to complete one for Silent Guard as part of onboarding.
Is AI allowed to make care decisions? CQC expects human oversight to be retained so professional judgement sits behind every decision. AI can inform decisions; a person should remain accountable for them.
Where can we read CQC's position ourselves? CQC's statement on AI in health and social care was published in May 2026, and national guidance for the responsible and safe use of AI followed on 29 July 2026. Both are freely available and worth reading before your next inspection.
CQC is not asking care homes to avoid AI. It is asking them to use it deliberately — with training, governance, data protection and human oversight behind it. Tools that keep a person in the loop and produce a factual record are the easiest to stand behind.
If you are reviewing the AI your home uses, one honest question is worth asking: for each tool, could you show an inspector who is accountable and where the human judgement sits? If you would like a call monitoring tool that is built around human oversight rather than automation, the 30-day Silent Guard pilot is a low-commitment way to see how it works. Start a pilot.
CareTime's Silent Guard is available now for a 30-day pilot. £49, 1-page pilot letter — exit by reply-email.
Join the 30-Day Pilot
CareTime