England's university regulator wants far more AI-free assessment, saying AI can complete unsupervised coursework

Students told researchers that some classmates finished assessments with AI in a few minutes and got similar grades.

By Think Facility · November 11, 2026 · UK AI policy

On November 11, 2026, the Office for Students, the regulator for higher education in England, published research it ran with Advance HE on how universities are coping with AI. The report is blunt about grading: when an assessment isn't supervised, students can often finish it to a high standard with AI tools and never show the knowledge it was meant to test.

The regulator wants more assessment that AI can't reach. Josh Fleming, its director of strategy and delivery, called in the press release for "far greater use of AI-free assessment."

Who: Office for Students and Advance HE
Report: Making AI work for students, OfS 2026.56
Published: 11 November 2026
Evidence: over 300 students and staff, from over 120 institutions
Research: between June and September 2026

What students and staff told the researchers

The work rests on roundtables and interviews, mostly with English institutions (so it says little about Scotland or Wales). Students described classmates who'd completed assessments with large language models, sometimes in a few minutes, and got about the same grade as people who'd done the work themselves.

Catching it is hard. The report says detection tools don't consistently and reliably identify AI use, and that it's usually impossible to establish for certain whether a student stuck to their university's AI policy. Students are often unclear what that policy allows in the first place.

Where assessments are unsupervised, it is frequently possible to complete them to a high standard using AI tools without demonstrating the knowledge and skills they are intended to test.

From Making AI work for students: reflections on institutional practice

Some universities have already moved. The redesigns the researchers saw were closed-book exams, presentations and other assessments held in controlled settings.

A change of tone since 2025

The regulator sounded different a year and a half ago. In a post on June 5, 2025, Fleming wrote that it regulates on principles partly to encourage experimentation, and that resisting AI would be "a temporary solution with significant opportunity costs" for institutions and students.

The November release still calls AI a huge opportunity, and the report still wants students to leave with AI skills. Fleming's language on the risk is much stronger now, though.

It also presents a generational risk to the system as it stands

Universities should help students develop AI skills while protecting confidence in degrees - Office for Students

The first thing the regulator says it'll do is require that assessments are valid and reliable measures of what students know. In many cases, it says, that will mean secure, AI-free assessment.

What happens to universities that don't change

The Office for Students says it'll hold institutions to account when they fall short. The release lists its levers: specific conditions of registration, monetary penalties, and suspending access to student support funding or its own grant funding.

The people it talked to didn't want new rules. Participants generally suggested that new conditions of registration on AI could be counterproductive and might constrain innovation, and some asked for clarity on how the existing ones apply. The regulator published guidance on exactly that alongside the report.

What it means for students and their degrees

Almost every undergraduate is already part of this. HEPI's Student Generative AI Survey 2026, from 1,054 full-time undergraduates, found 95% use AI in some form and 94% use generative AI to support assessed work. Fewer than half felt supported by their teaching staff.

For a student, the change is likely to show up in how they're graded. The examples the regulator gives of assessment designed to prevent AI use are closed-book exams, presentations, practical assessments and performances, and it says these are likely to give the most valid measure in many cases.

The report names what's at stake if universities don't adapt: the credibility of awards and the reputation of the sector. It also says some uses of AI replace the productive struggle that builds knowledge, which it calls cognitive offloading.

Whether AI improves or weakens learning depends on how it is used.

From Making AI work for students: reflections on institutional practice

It isn't all warning. The report says AI may bring particular benefits for students who are neurodivergent, have a disability or don't have English as a first language. And students in the research said they'd welcome more chances to build AI skills for their careers.

Advance HE says it'll launch an AI-focused Change Academy for its member institutions early in 2027.