On 11 November 2026 England’s Office for Students (OfS) and Advance HE published joint research arguing that unsupervised assessments can now often be completed to a high standard with AI, and the OfS said it will require institutions to make sure assessments validly and reliably measure knowledge and skills, often through secure, AI-free formats. It said it can hold providers to account where they fall short.
What the OfS announced on 11 November 2026
The announcement, published on the OfS website, accompanies a report titled Making AI work for students: reflections on institutional practice (OfS 2026.56). The OfS says universities and colleges have a key role in preparing students with AI skills while protecting the credibility of degrees.
Its central finding on assessment is that AI poses continuing risks to the validity of assessment and the credibility of qualifications, because some unsupervised assessments can be completed to a high standard with AI and little student engagement. The OfS says assessments designed to limit AI use, such as closed-book exams, presentations and practical assessments, are likely to be the most valid measures of attainment.
Josh Fleming, the OfS Director of Strategy and Delivery, is quoted in the announcement saying institutions need to make far greater efforts to address these risks, including far greater use of AI-free assessment.
What the OfS says it will do
According to the announcement, the OfS commits to:
- require institutions to ensure assessments validly and reliably measure knowledge and skills, often through secure, AI-free assessment, and to keep courses current;
- hold institutions to account where they fall short, using tools such as licence conditions, monetary penalties, or suspending access to student support and OfS grant funding;
- clarify how its existing requirements apply to AI and share good practice; and
- carry out further research into the scale and nature of AI’s impact on higher education.
The report page adds that participants generally did not support new OfS conditions of registration on AI and preferred clearer guidance on how existing conditions apply. The OfS framing is therefore an application of existing quality and standards requirements to AI, not a new AI-specific rule.
The evidence behind it
The report draws on conversations with more than 300 students and staff at more than 120 institutions, mainly in English higher education, between June and September 2026, plus wider academic and policy literature. It covers teaching and learning, assessment, AI literacy and institution-wide strategy; it does not cover academic research or the future of higher education.
On assessment, the report page records four points:
- unsupervised assessments can often be completed to a high standard with AI without showing the intended knowledge or skills;
- AI detection tools do not reliably identify AI use;
- students are often unclear about what AI use is permitted, and it is usually not possible to confirm whether they followed institutional policy; and
- some institutions have already moved to closed-book exams, presentations and other supervised assessments.
What it means for take-home and unsupervised assessment
The report’s position on detection matters most for integrity offices. If detection tools are not reliable and compliance with an AI policy usually cannot be confirmed, then a policy that depends on detecting misuse cannot by itself assure that a mark reflects the student’s own attainment. The OfS response is to move assurance into the assessment design: secure formats where the student’s own knowledge can be observed. That is the report’s reading; the press materials do not set a percentage or a deadline for any such shift.
The report also says institutions can use innovative, real-world assessments that integrate AI only where they are carefully designed to preserve validity and reliability. Take-home work is therefore not ruled out, but the burden of showing it still measures the intended outcomes sits with the provider.
Implications for assessment and academic-integrity offices
- Assessment mapping. Offices will want to know which assessments in each programme are unsupervised and which are secure, since that is the distinction the OfS draws.
- Policy clarity. The report finds students struggle to understand AI policies. Clear, programme-level statements of permitted use address a documented weakness.
- Detection as evidence, not proof. Our guides on how AI detection works and AI detection accuracy in higher education cover why scores alone are weak evidence in misconduct cases.
- Policy trends. For how institutions have been rewriting rules, see how universities are updating academic integrity policy for AI writing tools.
- Tooling decisions. Budget-driven changes to detection software are covered in our report on universities dropping Turnitin.
What Advance HE is doing
Advance HE says it will launch an AI-focused Change Academy for members in early 2027, a peer and action-learning programme to help leaders develop institutional AI strategies, and will convene sector leaders on AI in teaching, leadership, governance and inclusion. Kathryn Harrison-Graves, its Deputy Chief Executive, is quoted saying senior leaders are well placed to help by setting a shared vision for AI.
What is not yet known
The announcement does not name any institution as under investigation, and the sources do not say how many providers currently rely mainly on unsupervised assessment. The OfS says further research is planned to establish the scale of change. The details of how it will judge assessment validity in individual regulatory cases will depend on its guidance on existing conditions.
Sources
- Office for Students, Universities should help students develop AI skills while protecting confidence in degrees, 11 November 2026 (checked 2026-11-14).
- Office for Students and Advance HE, Making AI work for students: reflections on institutional practice, OfS 2026.56, 11 November 2026 (checked 2026-11-14).
