Regulator positions on AI are rarely just about the technology. They show where authorities think the main risk lies: authenticity, security, validity, public confidence, procurement, human rights, and qualification design. The newer source set strengthens a clear pattern: regulators and policy bodies are increasingly treating AI as a stress test for assessment design, not just a misconduct issue. The strongest practical question for assessment teams is whether current rules still support defensible results when AI use is normal, cheap, and difficult to see.

Regulator positions on AI are the public stances taken by qualifications regulators, government departments, standards bodies, and policy institutions on how AI should or should not be used in education and assessment. In practice, these positions tell assessment organisations what kind of risk the system is trying to manage: authenticity, fairness, due process, human rights, or the integrity of qualifications.

Regulatory signals shape what assessment providers can credibly do, what evidence they need to show, and how much trust the public will place in results. If a regulator links AI misuse to coursework authenticity, the implication is bigger than misconduct management: it may force a rethink of whether some assessment formats can still support high-stakes claims. Human-rights-aware governance matters too, because AI used in assessment can affect dignity, access, due process, and the ability to challenge decisions. The European Commission’s ethical guidelines add an important education-system signal: AI and data use should be handled with explicit ethical consideration, practical advice, and emerging competencies rather than assumed to be benign.

- **Authenticity**: whether learner work genuinely reflects the learner’s own contribution. - **Assessment design**: the extent to which the format itself supports valid claims under realistic conditions. - **Public confidence**: trust that results mean what they claim to mean. - **Human-rights impact assessment**: a structured way to ask whether an AI system could affect rights, fairness, or accountability. - **Ethical guidelines**: official advice on responsible use, practical caution, and competency development. - **Control measures**: steps such as teacher checkpoints, source referencing, detection tools, or governance reviews, which may reduce risk but do not automatically resolve design weaknesses.

The source set points towards a cautious regulatory posture. The strongest reading is that AI should be treated as a stress test for assessment design, not just a new misconduct problem. Where coursework can no longer be delivered securely and authentically, regulators may be willing to consider whether that format still suits a high-stakes qualification. UNESCO's regulatory survey, the Turing Council of Europe note, and the European Commission ethical guidelines all reinforce that governance now sits within a wider international conversation about AI risk, rights, and responsible educational use. There is also a clear practical lesson: layered controls can help, but they need to be judged against the underlying assessment purpose. Teacher checkpoints, clearer referencing requirements, and detection tools may improve assurance in some settings, but they do not by themselves guarantee authenticity or validity. The stronger governance point is that assessment bodies should know which risk model they are working to, not just which tool is available. The Commission guidance adds a useful complement here by pointing readers towards ethical considerations and practical advice rather than simple tool enthusiasm.

The open question is not whether AI creates risk, but what kind of response is proportionate. One view is that stronger controls can preserve coursework and digital assessment. Another is that some formats are too exposed to support the claims being made for them, especially in high-stakes settings. The policy challenge is whether human-rights and regulatory frameworks can be translated into practical controls that assessment teams can actually use. The direction of travel also appears unsettled on on-screen assessment. The source suggests careful balancing of innovation against public confidence, screen-use concerns, and system stewardship rather than rapid adoption. The European Commission guidance is not assessment-specific, but it does show a policy tilt towards explicit competence, ethics, and practical judgement rather than default deployment.

- **Validity risk**: a qualification may claim more than the evidence can support if authenticity is weak. - **Security risk**: AI may increase the ease of unauthorised assistance in coursework or home-based tasks. - **Rights and due-process risk**: decisions may be hard to challenge if the assessment body cannot explain its controls. - **Reputational risk**: public trust can fall if assessment formats appear easy to game. - **Procurement risk**: suppliers may overstate what detection or control tools can prove. - **Design risk**: treating misconduct as the only issue can leave a fragile assessment model unchanged.

A useful frame is to ask what problem the regulator is really signalling: - learner misconduct, - weak authenticity controls, - an over-exposed assessment format, - human-rights and due-process concerns, - or a broader confidence problem in the qualification system. That distinction matters because different problems call for different responses. Teacher checkpoints and source referencing may help in some contexts, but they do not automatically make a fragile coursework model secure. Digital delivery may improve consistency while still raising confidence and accessibility questions. Human-rights risk and impact assessment can be a useful way to make those trade-offs explicit before deployment. Ethical education guidance also supports a practical, not purely prohibitive, approach to AI competence and judgement. Assessment teams should ask what evidence would justify retaining coursework where AI misuse is plausible, and what would count as adequate support for moving towards on-screen assessment. They should also ask whether their governance framework is aligned with wider AI regulation, not just local policy.

| Option | What it means | Main upside | Main concern | |---|---|---|---| | **Keep current format with tighter controls** | Retain coursework or digital assessment but add checkpoints, referencing, moderation, or detection | Lower disruption; preserves established qualification structures | Controls may not be enough if the format is structurally vulnerable | | **Redesign the assessment** | Change the task so learner judgement is more visible or AI use is explicitly permitted | Better alignment between construct and evidence | Requires time, staff capability, and stakeholder agreement | | **Retire the format** | Remove non-exam assessment or limit digital rollout where risk is too high | Strongest simplicity when authenticity cannot be defended | Can reduce richness of evidence and increase exam burden |

A qualification team is reviewing coursework after repeated AI misuse concerns. Rather than jumping straight to detection, it asks whether the format still supports the claim being made about learner performance and whether a redesign could make learner judgement visible. If the evidence is weak, the decision may be to reduce dependence on coursework, add a viva, or move part of the task on screen only where that preserves the construct.

- FE Week / Schools Week reporting on Ian Bauckham, Ofqual, coursework authenticity, and digital exam policy. - UNESCO consultation paper on AI regulation. - Turing Institute / Council of Europe note on HUDERIA and human-rights risk and impact assessment for AI systems. - European Commission ethical guidelines for using AI and data in teaching and learning.

Vendor material in this area often frames AI detection, checkpointing, or digital assessment tools as straightforward mitigations. That is useful as a market signal, but it does not settle the regulatory question. The more important issue is whether the control meaningfully improves authenticity, supports due process, and fits the qualification claim.

### What is a regulator position on AI in assessment? It is the stance an authority takes on where AI creates risk and what kinds of controls or redesigns are acceptable. In practice, it often reveals whether the regulator sees the main issue as misconduct, design weakness, rights, or public trust. ### Does AI mean coursework cannot be trusted? Not necessarily. The stronger point is that AI makes authenticity harder to assume. Whether coursework remains viable depends on the qualification, the controls around it, and the strength of the evidence supporting its use. ### Is AI detection enough to protect assessment integrity? The evidence suggested here is no. Detection may be part of a wider approach, but it does not on its own resolve authenticity, rights, or validity concerns. ### Why do on-screen assessments come into this? Because regulatory thinking about AI sits alongside broader questions about digital delivery, public confidence, and the practical stewardship of the qualification system. The source suggests caution rather than simple enthusiasm for format change.

Tim Burnett (Admin)

Test Community Network. "Regulator positions." TCN AI & Assessment Wiki. Last reviewed 2026-05-02. https://www.testcommunity.network/wiki/regulator-positions.html

- FE Week / Schools Week article on Ian Bauckham, Ofqual, coursework authenticity, exam-board responses, and digital exam policy. - UNESCO consultation paper on AI regulation. - Turing Institute news on Council of Europe adoption of HUDERIA. - European Commission ethical guidelines on AI and data in teaching and learning.