AI detectors can turn academic integrity into a student trust problem
Updated Jun 28, 2026
AI detectors promise certainty at exactly the moment universities feel least certain about authorship. They can also make students less certain about whether their own writing will be read fairly. That is why Stefanus Galang Ardana and Merry Christiana's Teaching in Higher Education paper, "AI detectors, student anxiety, and authorial alienation: a qualitative study of affective control", matters for UK universities reviewing AI policy, misconduct processes, and student voice evidence on GenAI. The paper suggests detector use is not only a technical issue. It is also a trust, language, and educational relationship issue.
Context and research question
Universities have moved quickly to deploy AI detection tools in the name of academic integrity. The attraction is obvious. A detector appears to offer a scalable way to sort suspicious work from acceptable work, especially when staff are under pressure to make policy operational. The problem is that students do not experience detector use as a neutral background check. They experience it through anxiety, uncertainty, and assumptions about whether the institution trusts them in the first place.
Ardana and Christiana examine that problem through a qualitative case study of six non-native English-speaking thesis writers in Indonesia. Using the theories of Gilles Deleuze and Sara Ahmed, the paper asks how AI detection tools shape students' emotional and educational experience. The setting is not UK-based, but the practical question transfers well because many UK institutions are now balancing AI governance with support for multilingual writers, dissertation supervision, and clearer academic integrity communication.
Key findings
The first finding is that AI detectors operate as a form of surveillance before any formal accusation is made. Students do not wait for a case to be opened. They start adjusting behaviour as soon as they believe a detector may misread them. That matters for universities because a tool can change writing practice, help-seeking, and confidence even when no misconduct process is triggered.
The second finding is that this surveillance can produce authorial alienation. Students described being pushed away from their own writing voice, not because they wanted to cheat, but because they were trying to avoid looking suspicious. In effect, the detector becomes part of the writing situation itself. Instead of asking "How do I express this idea well?", students may start asking "How do I phrase this so a tool does not flag me?"
The abstract captures the consequence sharply:
"algorithmic surveillance compels students toward authorial alienation"
The third finding is that detector use can reinforce linguistic injustice. Because the study focuses on non-native English-speaking thesis writers, it shows how AI detection anxiety can attach especially strongly to students who already feel exposed by language expectations. If polished phrasing, uneven drafting traces, or unfamiliar sentence patterns are seen as suspicious, multilingual students may feel that their writing is being judged through a narrower margin of safety than others. For UK higher education teams, that is a serious warning when detector policies affect international students, home multilingual students, and anyone receiving language support.
The fourth finding is that detector-led governance can weaken teaching relationships. The authors argue that overreliance on these tools can create what they call "pedagogical abdication" by lecturers. In practice, that means a shift from dialogue and judgement towards tool-mediated suspicion. That sits alongside wider concerns about AI detectors, privacy, and false positives: when students think a tool is standing in for academic judgement, trust in the whole process can fall quickly.
Practical implications
For UK universities, the first implication is to treat detector outputs as weak signals, not decisive evidence. If a tool flags a piece of work, that should trigger human review, context gathering, and a chance for the student to explain drafting process, supervision history, and legitimate support used. The benefit is more proportionate academic integrity practice and less risk of turning uncertainty into accusation.
Second, institutions should ask students how detector use feels in practice, not only whether they understand the policy. Add open-text prompts that invite comments on fear of false accusation, confidence in staff judgement, perceived fairness for multilingual writers, and whether students feel safe disclosing how they worked. This complements the wider UK sector signal in Advance HE's cross-university evidence on student experiences of GenAI. The benefit is earlier evidence on trust before concerns harden into silence or disengagement.
Third, universities should separate AI-related comment themes more carefully. Policy confusion, detector anxiety, linguistic fairness, and trust in feedback are not the same issue. This is where Student Voice Analytics fits naturally: it helps institutions compare those themes across module evaluations, dissertation feedback, and local AI pulse surveys. A governed workflow such as our NSS open-text analysis methodology makes that evidence easier to interpret consistently. The benefit is clearer diagnosis, which makes policy revision more credible and more useful.
Finally, institutions should make follow-through visible once students raise AI-related concerns. If universities want honest comments on detector use, they need a clear route from concern to response, including small-cohort safeguards, escalation rules, and communication about what changed. That is why a robust student comment analysis governance checklist matters here. The benefit is not only cleaner governance. It is a stronger signal to students that raising a concern will lead to considered judgement rather than automated suspicion.
FAQ
Q: How should a university review its use of AI detectors after reading this paper?
A: Start by auditing where detector outputs currently sit in your process. If they function as near-automatic evidence, pull them back to a review-and-context stage. Then add one student feedback question on whether detector use feels fair, and one open-text prompt on what makes the process feel trustworthy or risky. That gives assessment and quality teams something more actionable than a technical accuracy claim alone.
Q: What should UK institutions keep in mind before generalising from this study?
A: This is a qualitative case study of six thesis writers in Indonesia, so it is designed to explain mechanisms, not measure prevalence. Its strength lies in showing how detector use can shape behaviour, trust, and language anxiety before a formal case exists. UK teams should therefore use it as a strong interpretive lens for local feedback and policy review, especially where multilingual writing support and dissertation supervision are involved.
Q: What does this change about student voice on AI more broadly?
A: It suggests student voice on AI should include surveillance, fairness, and confidence in institutional judgement, not only tool usefulness or policy awareness. Students may comply outwardly while still feeling watched, misunderstood, or discouraged from speaking honestly about how they work. Open comments are especially valuable because they reveal those quieter trust signals before they disappear into formal complaints or non-response.
References
[Paper Source]: Stefanus Galang Ardana, Merry Christiana "AI detectors, student anxiety, and authorial alienation: a qualitative study of affective control" DOI: 10.1080/13562517.2026.2643825
Request a walkthrough
Book a free Student Voice Analytics demo
See all-comment coverage, sector benchmarks, and reporting designed for OfS quality and NSS requirements.
- All-comment coverage with HE-tuned taxonomy and sentiment.
- Versioned outputs with TEF-ready reporting.
- Benchmarks and BI-ready exports for boards and Senate.
UK-hosted · No public LLM APIs · Same-day turnaround