Aipedia: Social workers

AI errors in records and fitness to practise

Aipedia, by the Apex broking team · Last reviewed 2026-10-30

An inaccurate AI-generated record can become a fitness-to-practise concern if a social worker approved it without proper checking. Research suggests frontline social workers carry much of the risk when AI tools get things wrong.

Part of: AI in social work

In short

Social Work England's research says social workers should be responsible for the accuracy and reliability of their case recordings and how they use AI. Standard 3.11 requires accurate records that show how decisions were reached, and 2.1 requires honesty. The Ada Lovelace Institute found that social workers may bear a disproportionate share of the risk from AI errors, while accountability between worker, manager and software maker is often unclear. No fitness-to-practise decision turning on AI had been published at the time of writing. The best protection is visible checking, honest correction and use of supervision.

How an AI error becomes a regulatory concern

Fitness-to-practise concerns usually arise when a pattern of poor practice, a serious incident or a question of honesty comes to light. An AI error can feed any of these. A transcription tool that misattributes a statement, a drafting tool that invents history, or a summary that omits a risk becomes the social worker's record once they approve it. SWE's research is direct: "Social workers should be responsible for the accuracy and reliability of their case recordings and how they use AI."

No SWE fitness-to-practise decision or UK court judgment turning on a social worker's use of AI had been published when this page was written. Any case would be decided under the existing standards.

The standards in play

The Social Work England standards most often in play are:

3.11: keep clear, accurate, legible and up-to-date records that show how decisions were reached.
2.1: be open, honest, reliable and fair. Concealing or explaining away an error is more serious than the error.
2.6: keep confidential information within the law, relevant where public tools were used.
3.1: use professional judgement rather than deferring to a tool.
5.6: do not use technology unlawfully or unethically.
6.1 to 6.5: raise concerns about unsafe practice and organisational wrongdoing.

See the professional standards.

Why the risk falls on frontline workers

The Ada Lovelace Institute interviewed 39 social workers in 17 councils about AI transcription. As reported by LocalGov, it found time savings, but wide variation in views on accuracy and oversight, and concluded that social workers may bear a disproportionate share of the risk from errors in AI-generated content. SWE's research also found that accountability between social worker, line manager and software maker is often unclear, even though employers should procure responsibly, set policies, quality-assure outputs and supervise.

In practice, the employer chooses the tool, but the individual signs the record. When something goes wrong, the signature is the first place anyone looks.

What can go wrong

Illustrative scenario: over several months, a social worker approves AI-drafted case notes with minimal checking because of caseload pressure. A family complains that notes record visits differently from how they happened. An audit finds repeated inaccuracies. The employer refers the matter to SWE. The social worker explains the tool and the workload, which is relevant context, but the concern is whether they kept accurate records under 3.11.

Protecting yourself

These habits help protect you:

Check visibly. Keep your own brief notes and compare them with AI drafts. Keep a habit of editing, not just approving.
Correct honestly. If you find an error in a saved record, correct it openly in line with your employer's procedure.
Use supervision. Raise concerns about tool accuracy and the time checking takes. BASW's guidance, as reported, warns checking may take longer than expected.
Raise concerns formally if a tool or workload makes accurate recording unsafe, under standards 6.1 to 6.5.
Use approved tools only. Unapproved use weakens your position.

Cover for regulatory proceedings

Employer arrangements usually deal with claims against the employer, but may not extend to an individual's defence before the regulator, so find out what support is available to you. Independent social workers should check whether their PI includes regulatory defence costs; see regulatory and disciplinary costs cover, notifying circumstances and independent social workers: does PI cover AI?.

How Apex helps

Apex arranges PI for independent social workers, including cover options for regulatory defence costs where available. A broker can explain how a policy would respond to an SWE investigation; see the AI in social work hub.

Sources

Frequently asked

Can I be referred to Social Work England for an AI error?

An error in itself may not be enough, but a pattern of inaccurate records, a serious incident or a question of honesty could lead to a referral. SWE's research says social workers should be responsible for the accuracy of their recordings and how they use AI.

Does it help that my employer chose the tool?

It is relevant context, and SWE's research says employers should procure responsibly, quality-assure and supervise. But standard 3.11 applies to the record you approved, so you still need to show you checked it.

What should I do if I find an AI error in a record I approved?

Correct it openly through your employer's procedure, tell your manager if it may have affected decisions, and make sure anyone who relied on it knows. Standard 2.1 requires honesty, and prompt, open correction is far better than an error found later by someone else.

Has anyone faced fitness to practise action over AI yet?

No SWE fitness-to-practise decision turning on a social worker's use of AI had been published when this page was written. Concerns would be judged under the existing professional standards.

Check your PI covers how you use AI

Tell us about your firm and the work you do. A broker will review how your current PI treats AI-assisted work, or arrange new cover. No fee for a quote. Or call 0117 325 0027.

Start a PI quote Request a callback

Apex Insurance Brokers Limited is authorised and regulated by the Financial Conduct Authority. Registered in England and Wales, company number 07014570. This page is general information about how AI is used in the profession and how insurance may respond. It is not legal, regulatory or insurance advice on your circumstances, and it does not guarantee that cover will be available or on what terms.