The Incident at NTU: Zero Marks for Suspected AI Use
In the bustling academic environment of Nanyang Technological University (NTU) in Singapore, a controversy erupted in early 2025 involving three students from the School of Social Sciences. These undergraduates, enrolled in a module exploring health, disease outbreaks, and politics, submitted a significant essay assignment that carried up to 45 percent of their overall module grade. The instructor, Assistant Professor Sabrina Luk, had explicitly prohibited the use of generative artificial intelligence (GenAI) tools such as ChatGPT, emphasizing the need for original research, independent thinking, and accurate citation practices. Briefing slides clearly warned: 'The use of ChatGPT and other AI tools are not allowed in the development or generation of the essay proposal or the long essay. You will receive a zero mark for the assignment if you are caught using ChatGPT and other AI for writing assignments.'
The essays flagged concerns when examiners discovered non-existent academic references, fabricated statistics, and broken web links—common hallmarks of GenAI 'hallucinations,' where tools generate plausible but inaccurate information. An investigation launched in April 2025 led to formal penalties in early May: zero marks, academic warnings, and grade point average impacts for all three students. This sparked public debate, amplified by Reddit posts from affected students detailing their side of the story.
Students' Perspectives and Initial Defenses
The students, all from public policy and global affairs, presented varied accounts. One Year 3 student vehemently denied generating content with GenAI, claiming she only used a reference organizer like Study Crumb or Citation Machine to alphabetize citations—a tool she viewed as innocuous. She provided a time-lapse video from Draftback, a Google Chrome extension tracking keystrokes, to demonstrate manual writing. Despite this, she received a permanent warning and GPA drop, feeling the accusation stemmed solely from citation errors common among undergraduates.
A Year 4 student admitted consulting ChatGPT for background research but insisted no generated text entered her essay, sharing her chat history during proceedings. She was unaware research use violated the ban and felt her transparency backfired. The third student similarly disputed full GenAI involvement, noting disclosure of limited AI for research. Their Reddit posts highlighted frustration over perceived shifting goalposts—from AI detection to general writing standards—and lack of formal hearings initially, fueling discussions on fairness in Singapore's high-stakes meritocratic education system.
NTU's Investigation and Academic Integrity Framework
NTU's response adhered to its structured academic misconduct process. Preliminary inquiries in April allowed students to explain. Two underwent face-to-face consultations in June 2025. The second student's appeal was swiftly rejected due to admitted GenAI use. The first formally appealed, prompting deeper review. NTU's general policy permits GenAI in assignments if declared, with users responsible for accuracy and proper citation. However, instructors can impose bans for skill-building tasks, communicating via instructions—a practice followed here to hone research abilities.
The university stressed academic integrity as foundational, where false citations undermine scholarly trust. Detection relied on manual review spotting implausible sources, not just AI detectors, reflecting Singapore universities' cautious approach amid low reported cases.
The Appeal Panel: Composition and Review Process
Responding to the formal appeal, NTU convened a specialized panel comprising senior academics and artificial intelligence experts. This step-by-step process included:
- Paragraph-by-paragraph essay analysis.
- Student presentation of writing process, including tools like Draftback.
- Demonstration of reference organizer use.
- Expert assessment of potential GenAI markers versus human errors.
The panel's rigorous scrutiny aimed for transparency, aligning with NTU's commitment to fair appeals while upholding standards.
Panel Findings: 14 Instances of False Citations
In July 2025, the panel upheld the zero mark, identifying 14 instances of false citations or data—not typographical errors but non-existent sources, misspelled author names (e.g., 'Lee' for 'Li'), incorrect dates, and fabricated statistics. These errors, often GenAI signatures, constituted serious misconduct as citation accuracy is core to academic research. The outcome was final, recorded internally. NTU urged respect for the process, noting no expulsions but emphasizing documentation's gravity. This shifted focus from initial AI accusation to verifiable inaccuracies, validating the penalty despite student protests.
For deeper insight into NTU's guidelines, visit their official AI assessment policies.
NTU's Generative AI Policy: Balancing Innovation and Integrity
NTU's framework defines GenAI (tools like ChatGPT generating human-like text/images/code) as permissible with caveats:
- Declaration required: State usage and method in submissions.
- Accuracy accountability: Verify all outputs; hallucinations demand correction.
- Instructor discretion: Bans for specific tasks to foster skills.
- Sanctions scale: Zero marks to expulsion based on severity.
This mirrors Singapore's proactive stance, where autonomous universities like NTU, National University of Singapore (NUS), and Singapore Management University (SMU) integrate AI ethically. SMU reports fewer than five cases in three years, underscoring rarity but vigilance.
AI Misconduct Across Singapore Universities: Low Incidence, High Stakes
Statistics reveal minimal cases: NTU's incident stands out, with no 2026 escalations reported. NUS and SMU echo low numbers, prioritizing education over punishment. A 2026 Channel NewsAsia report notes professors deeming AI detection a 'lost cause,' advocating process-based assessments like oral defenses or drafts. Singapore's Ministry of Education supports AI literacy, positioning universities as leaders in ethical tech use amid global cheating concerns.
Comparative table of policies:
| University | GenAI Stance | Reported Cases (2023-2026) |
|---|---|---|
| NTU | Allowed with declaration; bans possible | Handful, including 2025 case |
| NUS | Similar, emphasis on citation | Low |
| SMU | Less than 5 in 3 years | Minimal |
Challenges in AI Detection and Hallucination Risks
GenAI hallucinations—fabricated facts—pose detection hurdles. Tools like Turnitin flag patterns but falter on sophisticated use. NTU's manual review excelled here, spotting 14 anomalies. Experts recommend hybrid strategies:
- Process portfolios (drafts, notes).
- Live defenses.
- AI-proof prompts emphasizing personal insight.
In Singapore's context, where academic excellence drives careers, such breaches risk employability. A 2026 Straits Times analysis highlights rising AI tools but stagnant misconduct rates due to literacy efforts.
NTU's Forward Leap: Mandatory AI Literacy from 2026
Post-incident, NTU mandates AI literacy for all undergraduates from August 2026 via 'Science and Technology for Humanity'—a compulsory course covering ethical use, verification, and critical evaluation. Free premium Google AI tools access empowers responsible innovation. This aligns with Singapore's Smart Nation vision, training 100,000+ in AI annually. NUS and SMU follow suit, embedding modules. Details at NTU's AI empowerment initiative.
Stakeholder Perspectives: Faculty, Students, and Experts
Instructors like Prof. Luk prioritize core skills amid AI ubiquity. Students decry opacity, advocating clearer bans and hearings. Experts, including NUS Prof. Mohan Kankanhalli, call for policy evolution: 'Shift from prohibition to proficiency.' Parents worry about records impacting scholarships/jobs. Balanced views emphasize mutual responsibility—declare, verify, innovate ethically.
Photo by TSquared Lab on Unsplash
Future Outlook: Reshaping Assessments in Singapore Higher Education
By 2026, expect AI-integrated curricula, watermarking tech, and blockchain provenance. Implications include bolstered integrity, skilled graduates for Singapore's AI economy (projected SGD 22.4 billion GDP boost by 2027). Actionable insights for students:
- Always check instructor guidelines.
- Use GenAI as aid, not crutch—verify outputs.
- Build process evidence (drafts, videos).
- Leverage literacy programs proactively.
This case catalyzes positive change, ensuring Singapore universities lead in AI-era education.
Explore faculty opportunities at Singapore university jobs.