OfS says universities must use assessments that show what students know

The Office for Students says universities must ensure their assessments reliably show students’ knowledge and skills, with secure, AI-free assessment needed in many cases. Its response follows research finding that AI is reshaping higher education while many institutions still lack a strategic approach.

Watch Desk analysis

What happened

The research from the OfS and Advance HE found that some unsupervised assessments can be completed to a high standard with AI, without students engaging meaningfully with the work. Students and staff also said institutional AI policies can be hard to understand, and that staff training is often limited and voluntary.

University World News reports that the OfS will hold institutions to account if they fall short. Its regulatory options include specific conditions of registration, monetary penalties and suspending access to student support or grant funding. The regulator says it will also clarify how its existing requirements apply to AI and share examples of effective practice.

Advance HE plans to launch a peer-learning Change Academy for member institutions in early 2027.

Why it matters

This is more than advice to review assessment design. The OfS has connected reliable assessment to its regulatory powers, while leaving room for institutions to decide how best to test learning. Secure exams, presentations and practical assessments are among the approaches discussed in the research, not a universal new rule for every course.

The findings also point to a broader problem: students want clearer guidance on AI, while staff need time and training to make policies work. A rule that is clear on paper but opaque in the seminar room would be a familiar sort of progress.

Our read

The useful shift is from treating AI chiefly as a detection problem to asking whether an assessment demonstrates learning in the first place. That is a sturdier question than trying to catch every tool, though secure assessment should be used where it serves the learning goal, not as a reflex.

Universities now face both a practical expectation and a regulatory one. The detail to watch is how the OfS applies its existing requirements, and what evidence it expects when it says an assessment is valid and reliable.

What to watch

• How the OfS explains its requirements and what counts as a secure assessment.
• Whether enforcement action follows, and what evidence it cites.
• Whether Advance HE’s 2027 programme leads to clearer institutional AI policies.

Discussion spark: Should universities make supervised, AI-free assessment the default for proving learning, or should students be assessed on how well they use AI?

Watch Desk is operated by WittyWires as an independent cross-cutting AI news tracker. It does not speak for the organisations or people it covers.