The NTU Generative AI Incident: Unpacking the Accusations
In mid-2025, Nanyang Technological University (NTU) in Singapore found itself at the center of a heated debate over academic integrity when three undergraduate students in the School of Social Sciences were penalized for alleged misuse of generative artificial intelligence (GenAI) tools in a key assignment. The module, focusing on health, disease outbreaks, and politics, required a long essay worth 45 percent of the overall grade. The instructor, Assistant Professor Sabrina Luk, had explicitly prohibited the use of tools like ChatGPT from the outset, reinforcing this through multiple briefings and slides stating that any detected AI involvement would result in a zero mark.
The investigation began in April 2025 after submission, triggered by anomalies such as non-existent academic references, fabricated statistics, and broken web links—common hallmarks of GenAI 'hallucinations,' where tools generate plausible but inaccurate information. By early May, the students received formal sanctions: zero grades, permanent academic misconduct records, and grade point average (GPA) deductions. This case highlighted the tension between technological advancement and traditional academic standards in Singapore's competitive higher education landscape.
The incident gained traction through a viral Reddit post on June 19, 2025, where one student detailed how citation errors led to accusations of fraud, plummeting her GPA and adding a warning label that could affect future job prospects. Public discourse ensued, questioning whether universities were equipped to fairly adjudicate AI-related claims.
Student Defenses and the Appeals Process
The affected students, all in their third or fourth year pursuing public policy and global affairs, mounted robust defenses. One provided a Draftback time-lapse video demonstrating her writing process and argued that she used citationmachine.net merely for alphabetizing references, not content generation. Another admitted to minimal ChatGPT use for background research on COVID-19 statistics—unused in the final essay—and Citation Machine, emphasizing upfront disclosure and accepting responsibility for unchecked errors. The third claimed no GenAI involvement, only research aids, and alleged an unfair hearing where the professor dismissed evidence.
NTU offered consultations and appeals, including a S$40 formal review. Outcomes varied: one student's 'academic fraud' label was adjusted after proving a non-GenAI tool, but penalties stood. A pivotal appeal review panel, convened with AI experts, scrutinized one case in detail, identifying 14 instances of false citations or data not attributable to typos. The panel upheld the zero grade in July 2025, deeming citation of non-existent sources a grave breach of research integrity. Students expressed frustration over perceived lack of leniency and transparency, with one noting, 'I deeply regret not double-checking, but the process felt predetermined.'
NTU's Stance and GenAI Policy Framework
NTU maintains a pragmatic approach to GenAI, allowing its use in most assignments provided students declare it, verify outputs for accuracy, and cite properly—treating undeclared reliance as plagiarism. However, instructors retain discretion to ban tools for assessments targeting core skills like independent research and critical thinking. In this module, the prohibition was clear, aimed at fostering originality amid AI's rise.
The university's guidelines, outlined in resources like the InsPIRE initiative for pedagogical excellence, stress responsible integration. For research, GenAI aids proposals and editing but cannot author works or fabricate data; images must be labeled, and sensitive data protected under PDPA. Violations range from grade penalties to expulsion. NTU's response emphasized preparing students for an AI-shaped future through ethical training, not outright rejection.NTU's detailed GenAI policy underscores accountability.
Comparative Policies Across Singapore Universities
Singapore's autonomous universities align on balanced GenAI adoption. NUS permits use with disclosure, holding students accountable for content, and avoids AI detectors due to unreliability. SMU's framework guides ethical application, with faculty bans possible; it reported fewer than five misconduct cases in three years among 13,000 students. SUTD and SUSS note low but slightly rising incidents, prioritizing plagiarism checks over AI flags.
- NUS: Declare use; focus on value addition beyond AI output.
- SMU: Ethical checklists; cross-tool verification like Turnitin and GPTZero.
- SUSS: Heightened vigilance amid uptick.
A March 2024 report confirmed all six public universities endorse AI with honesty rules. By 2026, a new inter-university committee guides sector-wide strategies, reflecting Singapore's Smart Nation vision.
Photo by Markus Winkler on Unsplash
Challenges in AI Detection: False Positives and Hallucinations
Detecting GenAI misuse proves tricky. Tools like Turnitin flag non-native English or structured writing as AI (false positives up to notable rates), while hybrids evade scrutiny. In the NTU case, manual review of citations revealed issues detectors might miss. Experts like NTU's Eunice Tan note plagiarism platforms' unreliability, advocating contextual judgment—voice consistency, source validity, viva defenses.Professors call detection a 'lost cause,' urging redesigns.
GenAI hallucinations—fabricated facts—undermine trust, as seen with the students' erroneous references. Singapore unis adapt: in-person quizzes, process portfolios, AI critique tasks ensure understanding.
Consequences for Students and Academic Careers
Penalties extend beyond grades: GPA hits jeopardize honors, scholarships, jobs. Misconduct records linger, signaling fraud to employers valuing integrity. One student feared employability damage; another prioritized passing amid job offers. In Singapore's meritocratic system, where NTU grads dominate tech/finance, such stains amplify pressure.
Broader ripple: eroded trust between faculty-students, heightened anxiety. Yet, low stats—NTU 2-3 downgrades/semester, SMU handful over years—suggest rarity, but underreporting possible amid tool flaws.
Expert Insights and Adaptation Strategies
Educators advocate evolution. NUS's Aaron Danner pushes AI-inclusive assessments; Donn Koh mandates use for differentiation. SUSS's Grandee Lee tailors by skill level. NTU's InsPIRE offers GenAI resources, exemplars from global peers.
- Redesign: Iterative drafts, oral defenses, personalized prompts.
- Literacy: Verify outputs, prompt ethically.
- Ethics: Disclosure fosters transparency.
2026 developments: NTU mandates AI literacy for all from August, free Google tools; national committee coordinates.This addresses skills gaps.
Towards Ethical AI Integration in Singapore Higher Ed
Singapore positions as AI hub: Research Innovation Enterprise 2030 invests billions; NTU leads quantum-AI. Solutions blend bans with education: checklists (declare, verify, original input), AI-proof tasks (reflection essays, debates).
Student tips:
- Read instructor guidelines.
- Declare transparently.
- Cross-verify facts manually.
- Blend AI with personal insight.
Future Outlook: AI as Ally, Not Adversary
The NTU case underscores transition pains but signals progress. With low misconduct (no 2026 surges), focus shifts to empowerment: mandatory literacy, redesigned curricula prepare grads for AI-driven workplaces. Singapore's unis model global balance—innovation sans integrity compromise—ensuring students thrive ethically.
Stakeholders urge dialogue: faculty-student forums, policy updates. As GenAI evolves, so must higher ed, fostering thinkers who wield tools responsibly.
