0 of 19 GOC standards mention AI

Zero. That's how many of the General Optical Council's 19 Standards of Practice mention artificial intelligence, the document every registered optometrist and dispensing optician in the UK is measured against if a concern about their fitness to practise is ever raised. That would be a reasonable gap if nobody had thought to close it. Somebody did. The GOC ran a consultation in 2024 that specifically proposed strengthening the standards around the use of technology and AI when providing care. The revised standards that took effect on 1 January 2025, and remain in force today, still don't mention AI anywhere in their text.

If you're weighing up a diagnostic screening tool, an AI scheduling system, or an AI-assisted patient correspondence tool for a growing practice, that matters more than it sounds. It means the document that decides whether your use of that tool was reasonable was written to say nothing about it either way. This piece sets out what the 19 standards actually cover, what the GOC's own registrants say about their understanding of AI right now, what the two advisory bodies have published instead, and what to put in place while the statutory gap stays open.

What the 19 standards actually say about AI

The GOC's Standards of Practice for optometrists and dispensing opticians set out 19 standards of behaviour and performance (General Optical Council, Standards of Practice, current edition, direct page fetch September 2026). None of the 19 names AI, technology, or software specifically. The closest are generic: Standard 5 requires you to keep your knowledge and skills up to date, Standard 6 requires you to recognise and work within your limits of competence, Standard 7 covers conducting appropriate assessments and referrals, and Standard 8 requires adequate patient records. The preamble states that registrants are "professionally accountable and personally responsible" for their practice "no matter what direction or guidance" they're given, and must "always be able to justify" their decisions and actions.

That silence wasn't an oversight the GOC hasn't got round to fixing. A February 2024 consultation on revising the standards explicitly proposed strengthening provisions relating to "the care of patients in vulnerable circumstances and the use of technology and Artificial Intelligence (AI) when providing care" (College of Optometrists, coverage of the GOC consultation, February 2024). The standards that emerged from that process, in force since 1 January 2025, did add one new line to Standard 5: registrants must keep "updated on developments in digital technologies" and apply professional judgement "when utilising the data they generate" (GOC, "New GOC Standards launched," 1 January 2025). That's the closest the final document gets. The words "AI" and "artificial intelligence," proposed in the consultation, don't appear anywhere in the 19 standards as published. GOC head of strategy, policy and standards Charlotte Urwin confirmed the practical effect directly in June 2026: registrants "are expected to meet these standards regardless of what tools they use," and are "personally responsible for their practice... regardless of what guidance they have been given" (Optometry Today, "How AI is changing optometry," 4 June 2026).

So what for you: the fitness-to-practise test you'd actually be measured against contains no AI carve-out and no AI-specific rule you could be found to have broken. That cuts both ways. You can't point to a missing standard as cover for a bad outcome, and the GOC can't point to a specific rule you failed to follow either. The accountability sits entirely in your own judgement and your own documentation of it.

The regulator's own survey shows why the gap matters now

The GOC published its 2026 Registrant Workforce and Perceptions Survey on 3 September 2026, based on 3,451 responses collected between March and April 2026. Forty percent of registrants rated their knowledge and understanding of AI in optical care as good. Sixty percent rated it as poor. AI was most commonly used to help maintain knowledge and skills (12%), support diagnosis (8%), or write patient correspondence (8%). Only 22% of respondents had completed AI training in the previous 12 months. On concerns, 77% cited the risk of AI making errors, 76% cited uncertainty about legal responsibility if something goes wrong, and 68% cited a lack of transparency in how AI systems reach decisions (GOC, 2026 Registrant Workforce and Perceptions Survey, reported by Optometry Today, 3 September 2026).

A separate GOC dataset, its annual business registrant survey, shows the direction of travel: AI use among registrants rose from 5% in 2024 to 11% in 2025, with a further 28% saying they intend to use AI within the next two years (GOC annual business registrant survey, cited by Optometry Today, 4 June 2026). Whichever of the two surveys you weight more heavily, the pattern is the same. Adoption is climbing while training and self-rated understanding are not keeping pace, inside a workforce whose statutory standards say nothing AI-specific for that training to point back to.

So what for you: "everyone was using it and nobody had trained us properly" will not read as a defence if an AI-related concern ever reaches a fitness-to-practise process, because the accountability standard doesn't move for that. The gap between adoption and confidence is exactly the gap a practice's own policy needs to close, since the regulator's document currently doesn't.

Two bodies have moved faster. Neither is the regulator.

Two organisations have published more than the GOC has, and it's worth being precise about what each one is. The AOP, the membership association for optometrists (not the regulator), launched an AI and technology resource hub in 2026 with a glossary, FAQs, guidance on when an AI product counts as a medical device, and a five-question checklist for practitioners weighing up a new tool (AOP, AI and technology hub, aop.org.uk, checked September 2026). It's a genuinely useful working reference. It carries no statutory weight.

The College of Optometrists, a separate professional body focused on clinical standards, went further. It published an interim AI position statement in June 2025, developed with LOCSU and national optical bodies (LOCSU, 13 June 2025). In June 2026 it opened a six-week consultation on a fifth edition of its Guidance for Professional Practice, its first major revision in years, adding three new sections: artificial intelligence and new technologies, expanding scope of practice, and culture and governance. Clinical adviser Denise Voon described the update as reflecting "the rapid adoption of AI" in eye care (Optometry Today, 26 June 2026). The consultation closed on 3 August 2026, and the College expects to publish the new edition in winter 2026.

So what for you: don't plan around the College's new AI section as the moment this gets settled. Even once it's published, it will carry the same status College guidance always has, informative and best-practice, one level below the document your fitness to practise is actually tested against. The statutory gap this article opened with stays open regardless of what the College publishes in winter 2026.

Where this actually bites in a growing practice

None of this is abstract once you picture the workflow of a two or three-site optometry group. An autonomous diagnostic screening tool sits in a genuinely different risk category from an AI scheduling assistant, and only 8% of registrants report using AI to support diagnosis at all (GOC, 2026 Registrant Workforce and Perceptions Survey), which suggests most practices evaluating one right now have little internal precedent to lean on. We've covered the difference between autonomous and assistive diagnostic AI in eye care in detail, including which tools currently carry the regulatory clearance to make a call without a clinician reviewing it, in 92229: optometry's only autonomous AI code, a US-focused piece worth reading alongside this one if you're weighing a comparable tool here.

For scheduling, recall and patient correspondence AI, the operational risk is less about clinical error and more about consistency across sites. We've written before about how multi-site consolidation in optometry has already outpaced the admin layer meant to support it, in Optometry lost 10 points to consolidation. The same logic applies here: whichever optometrist or dispensing optician is on duty at a new third site needs the same AI use policy, the same review process, and the same documentation standard as the original site, not a version that's quietly drifted because nobody wrote it down centrally.

So what for you: treat "would this specific use of AI survive scrutiny under Standards 3, 5, 6, 7, 8 and 11" as its own question for every tool, asked and documented separately, rather than assumed to be covered by a general data-protection or IT sign-off.

What to put in place, in order

First: build an internal AI use policy mapped explicitly to the standards that already apply generically, valid consent (Standard 3), keeping knowledge and skills current (Standard 5), working within limits of competence (Standard 6), adequate patient records (Standard 8), and protecting patients from harm (Standard 11). Waiting for an AI-specific standard to arrive means waiting for a document that, as of September 2026, has no publication date.

Second: use the AOP's five-question checklist as a practical working framework now, while being clear internally that it's advisory guidance from a membership body, not a compliance requirement from the regulator.

Third: document the reasoning behind every AI-related decision as it's made, not reconstructed afterwards. The GOC's own wording, that registrants must "always be able to justify" their decisions, is already binding, and it doesn't require an AI-specific standard to apply.

Fourth: before extending any AI tool from one site to a second or third, confirm the same review process, the same named person responsible, and the same documentation standard travel with it. Treat this as a governance question, not just a technical rollout step.

The single most important thing to take from this

The GOC hasn't got this wrong so much as not got to it yet, and it said so itself: its April 2026 corporate update lists AI guidance as something it's still building the understanding to produce, with no date attached. Until that guidance exists, the accountability for how you use AI in practice sits with the 19 standards as written, general, tool-agnostic, and silent on AI by design rather than by accident. The College of Optometrists and the AOP have both published more than the regulator has, but neither carries the weight of the document your fitness to practise is actually tested against. For now, the connecting work between what AI tools can do and what the standards require is the practice's own, not the regulator's, and the evidence points toward writing that connection down before a vendor demo talks you into skipping the step. For the wider picture of how consolidation is reshaping multi-site optometry and where AI fits into that admin layer, see Optometry lost 10 points to consolidation. If AI governance turns out to be one gap among several worth checking, the AI Opportunity & Growth Assessment includes a standards-mapping review as part of its scope.