FCA authorised · FRN 724952 0117 325 0027Quote & buy →
Aipedia: Optometrists and opticians
The optometrist who relies on an AI reading of an OCT scan stays responsible for the clinical decision, because GOC Standard 7.8 requires professional judgement on technology-generated data. Late diagnosis of glaucoma or macular disease is a familiar optical claim, so how you use the output matters.
Part of: AI in Optometrists and opticians
In short
AI software can segment retinal layers, flag nerve fibre layer thinning, detect retinal fluid and suggest a likelihood of glaucoma or macular disease. Under the GOC Standards of Practice, effective 1 January 2025, the registrant must apply professional judgement to such data (7.8), work within scope (6.1) and keep clear, contemporaneous records (8.1). The College of Optometrists says AI is there to support clinicians, not replace them. Indemnity usually responds to a late-diagnosis negligence claim, subject to the wording, but a record showing the scan was reviewed, interpreted in context and acted on is what makes the claim defensible.
OCT may be offered in high-street practices, and some devices and cloud platforms add AI analysis on top of the standard printout. Typical outputs include automated segmentation of retinal layers, comparison with a normative database, flags for fluid or drusen, change analysis between visits and a summary risk indicator for glaucoma, age-related macular degeneration or other macular disease. Some tools run live in the test room; others process scans in a queue for later review, which can separate the AI result from the clinical decision in time.
The attraction is speed and consistency where many scans are taken. The risk is that a reassuring summary replaces the optometrist's own look at the B-scans, the optic disc, the pressures, the fields and the patient's history.
The relevant duties are:
The College's interim position on AI in eye care says AI is "there to support clinicians, not replace them". AI that analyses scans to support diagnosis may be an AI medical device regulated by the MHRA, so check the product's regulatory status before relying on it.
Illustrative scenario: a patient attends for a routine sight test with OCT. The software summary shows all parameters within normal limits. The optometrist records "OCT NAD" without reviewing the disc images in detail. Pressures are at the upper end of normal and a family history of glaucoma is noted but not connected to the scan. Some time later the patient is diagnosed with advanced glaucoma in hospital and alleges that earlier referral would have preserved more vision. The record shows only the software's verdict.
Other failure modes include false positives that lead to unnecessary referrals and patient anxiety, eyes outside the range of the normative database such as high myopes, poor-quality scans analysed as if they were reliable, and changes over time missed because the tool compared against the wrong baseline.
Late-diagnosis claims usually turn on two questions: would a reasonably competent optometrist have detected the problem and referred, and would earlier referral have changed the outcome. The defence depends heavily on the record. A note showing that you looked at the images yourself, considered risk factors, explained findings and gave safety-netting advice is far stronger than one that repeats the software result.
GOC Standard 12.2 requires adequate professional indemnity, with no minimum limit stated. A claim involving AI analysis is usually treated as a claim about the optometrist's judgement, and cover commonly responds on that basis, subject to the policy terms. Where a tool malfunctioned, the practice or chain that deployed it may also be pursued, and any recovery from the supplier depends on the contract. Read more in does optical indemnity cover AI errors? and the general guide to PI and AI mistakes.
Practical controls include:
Related: AI retinal photo screening, AI referral triage, AI scribes and records and the optical AI hub.
Apex arranges PI for optometrists and optical practices. A broker can check how your policy treats AI-assisted scan interpretation and whether practice and individual cover overlap; see our optometrist PI guide.
That is risky. Standard 7.8 requires you to apply your own professional judgement to technology-generated data. Review the images yourself, interpret them with the rest of the examination and record your own conclusion, so the record shows a clinical decision rather than a software output.
A patient will usually claim against the optometrist and possibly the practice. Whether the supplier shares liability depends on the facts and the contract. Professionally, the registrant who relied on the output is expected to have exercised judgement, so the record of that judgement is central.
Over-referral rarely leads to negligence claims, but it can cause complaints, patient anxiety and pressure on hospital services. Working within scope means using your judgement on whether a flag is clinically significant, not referring every alert automatically.
Yes, as good practice. Standards 2 and 3 on communication and consent support explaining how their scan was assessed, and the College's draft guidance adds transparency with patients as an expectation. Explain that you reviewed the result yourself.
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 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.
