FCA authorised · FRN 724952 0117 325 0027Quote & buy →
Aipedia: Doctors and medical practices
When an AI decision-support tool contributes to a missed diagnosis, the doctor who made the decision is usually the first target of a claim. Indemnity for AI-assisted care may depend on showing that a human, not the tool, made the clinical call.
Part of: AI in doctors and medical practices
In short
Doctors use AI tools that suggest differential diagnoses, flag risk of conditions such as sepsis or cancer, and prompt investigations. The GMC says doctors must use professional judgement and work within their competence (GMP paras 3 and 13), and that developers and employers stay responsible for their parts of a tool. The main risk is automation bias: accepting a confident suggestion or being reassured by a missing alert. One defence organisation's published position offers help with AI only where a human remains the decision maker, and excludes product liability for the software itself.
Decision-support tools sit inside clinical systems or run as separate apps. Some generate a ranked list of possible diagnoses from symptoms and findings. Others calculate risk scores from the record and prompt the clinician to consider an investigation or referral. In secondary care, prediction tools may flag patients at risk of deterioration. The GMC's research found doctors valued the efficiency, but some spent extra time checking decision-support output.
The BMA principles add that doctors must be able to challenge AI decisions. Under the Data (Use and Access) Act 2025, significant decisions based solely on automated processing of health data carry extra conditions and a right to human intervention, which is another reason to keep a clinician in the loop.
Illustrative scenario: a patient attends with headache and vague neurological symptoms. The decision-support tool ranks a benign cause highly and does not suggest urgent imaging. The doctor, reassured, gives safety-netting advice but does not examine fully or refer. The patient returns days later with a serious condition that earlier imaging would probably have detected.
The legal question is usually whether the doctor met the standard of a reasonably competent clinician, not whether the software erred. A tool's silence is not a defence if the history should have prompted investigation. The reverse also happens: a doctor overrides a correct alert without recording why. Doctors in GMC research named over-reliance and deskilling as key risks, and biased training data can make tools less accurate for some patient groups.
For NHS care, missed-diagnosis claims are usually handled under CNST or CNSGP in the normal way. For private work, the doctor's defence organisation or medical malpractice policy usually responds, subject to its terms. One defence organisation's published underwriting position offers assistance with AI only where "a human remains as the decision maker", AI use is within the member's specialty and the member follows law and national guidance. Fully autonomous systems, product liability and system defects are excluded, and such benefits are discretionary. If the tool itself was faulty, recovery from the supplier is uncertain: one defence organisation has argued AI is not clearly a "product" under the Consumer Protection Act 1987. See AI hallucination liability.
See also AI in radiology, chatbots for clinical questions, does medical indemnity cover AI and the doctors AI hub.
Apex arranges indemnity for doctors and clinics, including private work outside NHS schemes. A broker reviews your cover and how your policy treats AI-assisted diagnosis; see does PI cover AI mistakes.
Not on its own. Claims usually turn on whether you met the standard of a reasonably competent clinician. Showing that you took a proper history, examined, considered the AI output critically and recorded your reasoning is far more useful than showing what the tool suggested.
Possibly the supplier, the deploying organisation and the clinician, depending on the facts. One defence organisation's published position on AI excludes product liability and system defects, and whether AI counts as a product under the Consumer Protection Act 1987 is disputed, so the clinician may be pursued first.
Yes, where it affects decisions about them. The GMC says doctors should discuss AI use with patients, including alternatives, uncertainties and limitations. The government has accepted a recommendation recognising patients' reasonable expectation of being told about AI use.
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.
