Aipedia: Doctors and medical practices
A GP practice that uses AI or automation to file results and letters remains responsible for any abnormal result that is never actioned. Missed results can lead to delayed diagnosis allegations against general practice.
Part of: AI in doctors and medical practices
In short
GP practices receive large volumes of lab results and hospital letters, and some now use AI or rule-based automation to file normal results, extract actions from letters and route documents to clinicians or administrators. The risk is an abnormal result or a request in a letter that is filed without anyone acting on it. The GMC expects supervision of delegated tasks (paras 61 and 66) and reporting of incidents involving software (para 73). Deploying organisations should have a DCB0160 safety case. Claims from NHS general practice are usually handled under CNSGP.
Automation in general practice usually takes three forms. Results tools file results within set ranges as normal, sometimes with an automatic message to the patient, and pass abnormal ones to a clinician. Correspondence tools read hospital letters, summarise them, code diagnoses and identify actions such as medication changes or follow-up tests. Workflow tools then route documents to administrators, pharmacists or GPs. Generative AI is increasingly used to summarise long letters for a quick clinical review.
Practice partners also need to consider the GMC's expectation that doctors raise concerns if a technology may put patients at risk.
Illustrative scenario: a hospital letter recommends a repeat scan in three months to check a small finding. The AI summary records the diagnosis but not the follow-up request, and the letter is filed by an administrator with no task created. Nobody arranges the scan. When the patient presents a year later, the finding has progressed and the family alleges the practice failed to act on the letter.
Other failures include results filed as normal because the reference range did not suit the patient, trends missed because each result was within range, and automatic messages telling a patient their result is normal when a clinician would have acted. Missed results often come to light only when harm has occurred, so they can involve several years of practice history.
Responsibility can also blur when work is shared. Practices in a primary care network may share pharmacists or document teams, and a supplier may update the tool's rules without the practice noticing. If the tool changes, the safety case and hazard log should be reviewed, and staff told what is different. NHS England's scribe guidance warns that liability may default to the provider where a supplier does not accept responsibility, so contracts should be clear about who fixes errors.
CNSGP covers clinical negligence in NHS general practice for incidents from 1 April 2019, with ELSGP covering earlier liabilities. A missed-result claim from NHS work is usually handled by NHS Resolution under those schemes, whatever system was used. The scope table excludes data protection breaches and third-party cyber incidents, so a misdirected automatic message or a data leak is not covered. Practices may also face a CQC inspection or a GMC concern about governance, which state schemes do not cover. Private GP work needs separate indemnity; see CNSGP and private work.
See also AI-drafted letters, CNST, CNSGP and AI, AI in prescribing and the doctors AI hub.
Apex arranges indemnity for doctors and clinics, including private work outside NHS schemes. A broker reviews your cover for gaps outside CNSGP, such as data incidents, and how your policy treats AI-assisted workflow; see cyber insurance and AI.
The practice and the clinicians responsible for the workflow. Delegating filing to software or administrators does not remove the duty to act on results. GMP paras 61 and 66 require proper supervision and training for anyone acting on routed documents.
Usually, for NHS general practice incidents from 1 April 2019, whatever system was used. CNSGP does not cover data protection breaches, cyber incidents or regulatory proceedings, and incidents before April 2019 fall under ELSGP.
NHS England expects deploying organisations to complete DCB0160, including a safety case, hazard log and monitoring, for AI scribes, and the government has accepted that guidance for other AI will follow that model. A safety case is good practice for any tool that routes clinical documents.
Tell us about your practice and the work you do. A broker will review how your current cover treats AI-assisted work, or arrange new cover. No fee for a quote. Or call 0117 325 0027.
Start an indemnity quote Request a callbackApex 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.