Using AI safely
Keep people and their information safe.
The rules, checks and paperwork that let you use AI without putting anyone at risk. Start with the one-minute check or the golden rules.
In short
Five things everyone should know
Never paste identifying details into an unapproved tool
Names, addresses, health details, photos. Treat unapproved tools as public.
A person checks everything
AI makes confident mistakes. Read and correct before anything is used.
AI drafts, people decide
No decisions about care, risk, eligibility or safeguarding.
Be open about it
People you support should know when AI is involved, before it starts rather than after.
Get it on paper
A short policy, an approved tool list and a DPIA for anything involving personal data.
Your data and AI tools
What happens to what you type, and how to choose a tool that handles it properly.
CQC, the law and your board
What regulators expect, and the paperwork that shows you have thought it through.
Policies and staff rules
Clear rules staff can remember and follow.
Spotting mistakes
The errors AI makes, and quick ways to catch them.
Safeguarding
What must never be automated, and the quiet risk of AI smoothing a pattern out of your records.
Telling the people you support
Explaining AI plainly, and being honest about what you can offer if someone objects.
Free AI policy builder
Answer questions about your organisation and get an AI use policy, a one-page staff summary and an action plan, in about fifteen minutes. It uses fixed wording from published guidance; gaps in your answers are marked, not filled in for you.
Free document anonymiser
Removes names, addresses and other personal details from a document, so you can use it with AI more safely. Always read the result: context can still identify someone.
Straight answers
Questions we hear from care leaders
- Can AI write our care plans?
- It can draft sections from notes you already hold. Someone who knows the person still writes the plan with them and checks each line.Drafting care plans
- Is it safe to put people's information into ChatGPT?
- Not into a free or personal account. Some business plans can be set up safely, with a data processing agreement and a DPIA.Free and paid tools
- Will using AI affect our CQC rating?
- CQC says the presence or absence of AI does not predict a rating, and that it does not assess or approve specific technologies. It has not yet said how inspectors will test this in practice.What CQC says
- Do we have to tell people we use AI?
- Yes. The ICO requires you to tell people before a new use of their data begins, not after. Where someone objects, be honest about what you can actually offer rather than promising an alternative you cannot provide.Telling people you support