England’s university regulator warns AI-weakened assessment could bring sanctions

Universities in England could face sanctions if their assessments fail to reliably measure students’ knowledge and skills amid widespread AI use. The Office for Students has called for more AI-free assessment, alongside a joint report with Advance HE published on November 11 that warns AI threatens confidence in degrees and students’ incentives to learn.

The research, conducted between June and September 2026, draws on conversations with more than 300 students and staff from over 120 institutions. It examines primarily English higher education, covering teaching, assessment and AI skills. Its findings combine concerns about submitted work with examples of institutions changing how they judge learning.

A good submission may not prove learning

The report identifies a basic weakness in unsupervised assessments: students can frequently use AI to produce high-standard work without demonstrating the knowledge or skills being tested. The risk extends beyond individual misconduct. If grades reward a finished answer rather than the student’s understanding, the report says, students have less incentive to engage in learning.

Detection offers no dependable fix. The report says AI-detection tools do not consistently and reliably identify AI use. Students are also frequently unclear about whether and how they may use AI in assessed work. Institutions usually cannot conclusively establish whether a student has followed their AI policy.

Students interviewed described peers finishing assessments with AI in minutes and receiving grades similar to those earned without it. Some recalled peers who could not remember basic details of work they had just submitted. Academics described batches of submissions with similar styles and references, according to the Guardian’s coverage of the findings.

We need to see far greater efforts to tackle these risks, including far greater use of AI-free assessment.

Josh Fleming, director of strategy and delivery at the Office for Students

Changing the evidence students must provide

Some institutions have already increased their use of closed-book exams, presentations and other assessments in controlled environments. The report says secure assessments designed to prevent AI use will often provide the strongest confidence that students are demonstrating their own abilities. It also calls for assessment approaches that are accessible as well as robust.

Assessment redesigns described in the findings

• In-person written exams and oral examinations, known as vivas, give institutions ways to assess students directly rather than relying solely on submitted coursework.
• Tasks that ask students to critique or edit AI-generated output make the machine’s response material for the student to examine.
• Process-based assessments ask students to explain how they researched and produced their work, adding an account of the work behind the final submission.

The report does not reject assessments that incorporate AI. It says approaches reflecting real-world applications can be positive, but require thoughtful design to preserve a reliable measure of students’ knowledge and skills. Allowing AI is not, by itself, a substitute for establishing what the student can do.

Teaching AI skills while protecting grades

The distinction also runs through the report’s account of studying. Students use AI to explain difficult concepts, summarise documents and generate practice questions. Used carefully, it can help identify gaps in understanding. Used to replace reading, research or problem-solving, it can bypass the effort needed to develop knowledge and skills.

Preparation remains uneven. Many students see effective AI use as important to future employment, yet reported limited training or none at all. The report says institutions need to account for students’ different starting points and unequal access to tools. Staff training is frequently optional and often attracts low participation.

Research participants generally warned that new AI-specific registration conditions could constrain innovation. Some instead sought clearer guidance on existing requirements. The Department for Education took a firmer regulatory line, saying it expected the OfS to enforce existing standards effectively and make its framework explicit, strong and durable on AI use.