Universities risk losing students’ trust over failing AI rules
Vendan Ananda Kumararajah
- Published
- Opinion & Analysis

New research reveals students are afraid to declare permitted AI use, while assessments struggle to distinguish genuine learning from machine-generated work. Universities must prove their policies are fair, effective and capable of correcting failures before confidence in degrees is undermined, says Vendan Ananda Kumararajah
Some students say they would avoid declaring their use of AI in assessed work even when the rules allow it. They fear the declaration could cast suspicion on their work and cost them marks. That finding, from research published last week, gives universities a way to test whether their AI policies work. If students are reluctant to report permitted use, what will the institutions change, and how will they know that students feel safer disclosing it?
The Office for Students (OfS), which regulates higher education in England, carried out the research with Advance HE. It draws on the views of 244 higher education staff and 91 students and student representatives from at least 127 institutions, with contributions from across the UK. Participants were self-selecting, so the findings cannot tell us how widespread these problems are. They do, however, show where universities can look for weaknesses in their own arrangements.
In February, I argued that students needed clear rules about what help they could use and lecturers needed institutional support to apply them. Some students in the new research described expectations that changed or varied across departments and staff. This exposes a practical weakness in disclosure-based governance. A declaration depends on students understanding what is allowed and trusting the way the rules are applied. Where students fear that declaring permitted use will count against them, institutions need to find out why before they ask for more disclosure.
Assessment presents a more fundamental problem. The report describes unsupervised tasks that AI can complete to a high standard without the student demonstrating the knowledge or skill being assessed. It also warns against putting too much weight on AI detection tools and relying on misconduct procedures to make up for weaknesses in how tasks are designed. Where an assignment can no longer establish what a student understands, responsibility extends to those who design, approve and retain it.
The University of Bath changed its approach after finding that specified uses of AI in one category of assessment could not be policed reliably and fairly. In September 2026 it separated assessments in which students have no access to AI from those designed on the assumption that students may use it. Bath emphasises that even an AI-free exam must test what the course was intended to teach. An exam can be secure and still test the wrong thing. An assignment that permits AI can be valuable if it shows what students understand and can do.
Bath shows that experience can change the rules. The follow-up is to establish whether the revised assessments are fairer and give lecturers a better basis for judging what students have learned. That will take evidence from running the assessments.

The report also describes the University of Kent’s rollout of a licensed AI platform in March 2026. Use is optional, and students who want access must complete training first. Kent shows how a university can offer a tool and prepare students to use it responsibly. Questions about access remain. Universities need to check whether students who cannot afford premium tools, or who choose to work without AI, can still participate fairly. They should also consider how changes to assessments affect accessibility.
The deeper governance question is what institutions do with what they discover. Effective correction requires an identifiable decision, someone responsible for carrying it out, and evidence that the response addressed the problem.
The same test applies when a university uses AI to give feedback. Suppose students repeatedly find errors in the comments they receive. Someone needs to have the authority to investigate and pause the use of the tool, and affected students need support. Before continuing, the university should be able to say what it has checked and what gives it confidence that students will receive accurate feedback.
The OfS has already set out how it will respond to the research, including clarifying how its requirements apply in the context of AI, sharing examples of effective practice and undertaking further research. Advance HE has announced plans to help member institutions develop their strategies through a programme in early 2027. Many participants asked for clearer guidance on existing requirements. Some expressed concern that additional regulation could constrain innovation. There is a credible case for that restraint. It becomes stronger when universities can demonstrate that existing arrangements allow them to act when a tool or assessment fails.
Universities should explain who will act on the findings, when progress will be reviewed and what changed as a result. If unclear rules, weak assessments and uneven support persist, students may come to think that doing well depends on the technology they can access or the risks they are willing to take. Universities ask students to substantiate their work. They should now apply the same discipline to their own AI decisions.

Vendan Ananda Kumararajah is an internationally recognised transformation architect and systems thinker. The originator of the A3 Model—a new-order cybernetic framework uniting ethics, distortion awareness, and agency in AI and governance—he bridges ancient Tamil philosophy with contemporary systems science. A Member of the Chartered Management Institute and author of Navigating Complexity and System Challenges: Foundations for the A3 Model (2025), Vendan is redefining how intelligence, governance, and ethics interconnect in an age of autonomous technologies.
READ MORE: It’s time to pull the plug on dangerous AI agents. An Anthropic AI agent tried to plant malicious code in software during a UK government safety test despite recognising the potential harm, according to a new report. The attack was thwarted but future attempts could succeed, exposing people and businesses to serious cyber threats. To prevent this, AI operators must urgently restrict agents’ access to external systems and withdraw it when dangerous behaviour emerges, says Vendan Ananda Kumararajah.
Do you have news to share or expertise to contribute? The European welcomes insights from business leaders and sector specialists. Get in touch with our editorial team to find out more.
Main Image: Kampus Production / Pexels
TOP STORIES
-
Meet the seven Anthropic founders worth $23bn each after record-breaking stock market debut -
World's robotics leaders to gather in Germany in race to put AI to work -
Britain's most expensive seaside resort dominates £2m property sales -
Beavers return to Loch Ness after 500 years -
Swiss data centre to heat homes using waste energy from AI servers -
Uzbekistan attracts $44.1bn in foreign investment and loans in nine months -
Serko raises revenue forecast as Booking.com business travel bookings surge -
Air India and Airbus unveil India's first corporate green aviation fuel scheme -
Best Western loyalty scheme to be scrapped in major hotel rewards shake-up -
First class delivery: China's postal service puts humanoid robots to work across eight cities -
AI set to transform university librarians' jobs as 95% predict surge in demand -
Washington Dulles unveils new concourse as $20bn airport expansion takes off -
The European Autumn 2026 edition – out now -
WATCH: Greenpeace drops giant David Attenborough boulder into North Sea – on site of £12bn windfarm -
Himalayan nations demand urgent action as UN warns of irreversible glacier losses -
6,000 march against ‘monstrous’ AI data centres as Scotland faces building boom -
Shop till Ai drop? AI agents set to buy for us as tech boss warns of retail revolution -
Specs appeal: Welsh Parliament bans smart glasses -
European motoring legends behind Mercedes, BMW and McLaren to enter Hall of Fame -
Putin 'plotting to break NATO without firing a shot', Latvian military analyst warns -
British AI chip startup hits £5bn valuation as government backs bid to challenge global tech giants -
World's biggest military airshow moves to ‘Britain’s Area 51’ after Iran war disruption at RAF Fairford -
Lisa Nandy threatens crackdown on entertainment industry after David Walliams allegations -
UK and European experts demand tougher AI laws as Australia considers criminal penalties after OpenAI breach -
BBC loses bid to throw out Trump's $10bn defamation lawsuit over doctored Panorama speech
Universities risk losing students’ trust over failing AI rules
Vendan Ananda Kumararajah
- Published
- Opinion & Analysis

TOP STORIES
-
Powerful men need to hear ‘no’ from people they can’t sack -
Universities risk losing students’ trust over failing AI rules -
007 reasons why James Bond should be played by a disabled actor -
Carrie Gracie's BBC pay battle exposes sexism's dirty secret -
The great AI shame game -
Businesses risk losing staff trust over poorly explained AI plans -
Monty Python filmmaker claims to have found the Garden of Eden -
The short and the short of it: children's fiction is shrinking -
Grandfathers can play vital role in preventing boys being brainwashed by the manosphere -
Landlord tax rises risk pushing rents higher -
Why Britain must put Black women at the heart of its public art -
Trumpism will survive the Democrats' midterm victory -
How Putin plans to break NATO without going to war -
The scuba diving project helping to protect Cornwall’s lobster population -
For Your Eyes Only. Why the new Bond needs a woman's perspective -
It's time to pull the plug on dangerous AI agents -
Sound of silence: Why AI could leave disabled musicians unheard -
Could the BBC survive without the licence fee? -
Visionaries in focus – how spectacles became symbols of political power -
The coach driver building a dating platform in his sixties -
If I were Chancellor, disabled people’s independence would be at the heart of the Budget -
Trump, Europe and the battle over who controls free speech -
Frozen bank accounts, NHS errors and food shortages. How AI could trigger a national crisis -
AI might kill us all, but not until it’s got what it wants -
UK government urged to ban personalised price rises based on shoppers’ online behaviour



















































