AI in adult social care: what the new DHSC guidance means for your service

Artificial intelligence is already in your care service, whether you planned it or not. A care worker drafting notes in a free chatbot. A manager using AI to write a rota policy. An acoustic sensor flagging a fall at 3am. On 29 September 2026, the Department of Health and Social Care published Using AI in adult social care, its dedicated guidance for providers.

The message is that AI can free up time for care, but only with the right safeguards and a human in charge.

Where AI is helping already

The guidance gives practical examples:

• Care planning: generative AI drafting care plans, assessments and daily records.
• Sensors: monitoring movement, sound and vital signs, and alerting staff to deterioration. Acoustic monitoring can reduce disruptive night-time checks.
• Pain assessment: facial recognition software supporting people who cannot easily report pain.
• Spotting patterns: in one example, incident data showed falls happening at the same place and time each day. The cause was sunlight, and fitting blinds reduced the falls.
• Administration: meeting notes, recruitment, onboarding and audit summaries.

The falls example is my favourite, because it shows AI at its best: noticing a pattern busy people miss, then handing the decision back to them.

The four rules that matter most

1. Never put personal data into free tools. The guidance says not to enter personally identifiable information into free or online AI tools without a contract that guarantees how it is used. Check where your data is processed, ideally within the European Economic Area.
2. A human always reviews the output. The guidance says to "always include a human review" of AI output, and for care plans a member of care staff must review the plan before it is used. AI can draft, but it cannot know Mrs Patel prefers her tea before her tablets.
3. Have an AI policy. The guidance asks every organisation to consider one, even if the policy is that staff may not use AI for work. It should say who is accountable, how oversight works and what checks are made, and all staff should see it.
4. Watch for bias and poor data. AI reflects the data it learns from. Inaccurate or incomplete records produce inaccurate outputs, and bias can creep in through training data, design or human labelling.

The care plan question

Care plans are where providers are most tempted and most exposed. AI can produce a tidy, well-written plan in seconds, but a plan that reads well is not the same as a plan that is personal. The guidance notes that evidence AI produces truly person-centred plans is still limited. Treat every AI draft as a starting point, and make sure the person and their family shape the final version.

Where to start

1. Find out what your staff already use. You may be surprised.
2. Write a one-page AI policy covering what is allowed, what is banned and who signs off.
3. Choose one low-risk use. Meeting notes or audit summaries are a sensible first step.
4. Train your team. The guidance highlights workforce upskilling and digital champions. Confidence comes from understanding, not from a ban.
5. Review and improve. Set measurable outcomes and check them.