AI cheating still rare in Singapore universities, but risks are growing

Somewhere on Reddit, a university student vents that they got zero marks because ChatGPT helped with an assignment. The comments flood in. Some call it deserved. Others shrug. “Everyone uses it,” one reply reads. “You just weren’t careful.”

Officially, Singapore’s universities report only “a handful” of AI-related academic misconduct cases. Unofficially, the vibe is different. Students talk about ChatGPT the way we talk about Wikipedia in the 2000s—ubiquitous, casual, low-stakes. Not a cheat code. Just part of the system. The real shift isn’t in who’s copying. It’s in how learning itself feels increasingly outsourced—and no one’s quite sure where the line is anymore.

Let’s start with the numbers. Singapore Management University (SMU) says it’s had “less than a handful” of AI-related misconduct cases in three years. SUTD and SUSS say the same. NTU won’t comment on overall numbers, but it made headlines for awarding three students zero marks after detecting AI-written assignments. And yes, that case only went public because a student posted about it online.

Across public universities, the official message is: we’re watching, we’re updating policies, and we want students to use AI responsibly. The unofficial message? Everyone’s already using it—and universities are scrambling to catch up not just in tech policy, but in pedagogy, culture, and credibility. Because the real issue isn’t whether AI is being “misused.” It’s what we define as learning, and who decides what counts.

Scroll through student group chats, Reddit threads, or Telegram circles and the story gets less pristine. One fourth-year law student at SUSS told The Straits Times: “It has become so rare to see people think on their own first before sending their assignments into ChatGPT.”

It sounds dramatic until you realize she’s not angry. She’s just...naming it. Other students admit to using AI for “non-core” modules, fixing grammar, summarizing notes, or generating research outlines. One NTU student called ChatGPT his “study buddy”—a smarter, faster version of Google and groupthink combined.

There’s no great academic conspiracy here. No villain monologue. Just students navigating overwhelming coursework, high expectations, and an education system built for another time. AI isn’t making them lazy. It’s making visible what’s already been fraying: attention, meaning, and a sense of ownership over their learning.

Let’s talk about vibes. Because what’s really happening isn’t about tools or policies. It’s about a vibe shift on campuses—and how students are adjusting to the dissonance between old rules and new realities.

In-person class? Laptop open, ChatGPT in one tab. Group project? Divide the prompts and feed them into Claude. Essay question? Generate a response, then rewrite just enough to sound “authentic.”

None of this feels shocking anymore. But it does feel hollow. Performative. Like school has become a theater where everyone pretends to care, while the real work gets done in side windows and offscreen tools. Students aren’t necessarily trying to cheat. But many are quietly opting out of the illusion that university assessments measure anything deeper than deliverables. And professors? Some know. Some don’t. Some are trying to fight it. Others are trying to design around it.

Here’s where things get sticky. Universities say they encourage “responsible use” of AI. That’s code for: tell us when you use it, and don’t let it do the whole job. But that message lands inconsistently across courses, departments, and faculty members.

At NUS, AI use is allowed in take-home assignments—if cited. At SMU, instructors tell students what tools are allowed, but only some teach how to prompt or fact-check. At SUTD, it’s built into the curriculum for design thinking. At SIT, it’s encouraged in advanced coding modules but restricted in basics. So students are left to interpret the norms like amateur policy analysts. Some over-disclose. Others quietly adapt. Because when the boundary between “tool” and “proxy” is this blurry, the safe move is to make the assignment look human—even if it’s not.

Academic dishonesty used to be easy to spot. Copy-paste a paragraph. Steal from a peer. Hire someone to write your essay. These were sins with clear fingerprints.

But now? The line between efficiency and dishonesty is blurred by design. And that blur is revealing something deeper: disengagement disguised as performance. A student who uses AI to generate an essay isn’t necessarily dishonest. They might just not care. They might not see the value in the task. Or they might be too overwhelmed to do it any other way.

And that’s the part institutions don’t want to confront. Because it’s not about tools. It’s about trust. About relevance. About the slow erosion of meaning beneath all the GPA chasing and skills checklisting. If students are outsourcing work not to cheat but because they don’t believe the work matters—that’s not a tech problem. That’s a design problem.

Not every student is tapping out. Some draw clear boundaries. A third-year SMU political science major said she uses AI to fix her grammar but never to write the whole essay. An NTU economics student calls it “a smart study buddy,” not a ghostwriter. A computer science major at SUTD uses it, but deliberately leaves in his mistakes—because he wants to learn from them. These students aren’t rare. They’re just quieter.

They still believe in the process. They want to stretch their ideas, not just optimize their outputs. But they’re operating in a landscape where efficiency often looks like wisdom—and slowing down feels like falling behind. Their restraint is not just personal ethics. It’s a quiet act of resistance against an educational system that often rewards speed, polish, and deliverables more than depth.

There are faculty members who see the whole picture. SMU’s Associate Professor Seshan Ramaswami lets students use AI—but only if they disclose how, and critique the output. He designs assignments that AI can’t do alone: local context, live debates, messy data. He even uses a chatbot in class—but tells students not to trust it blindly.

SUTD’s Dr Thijs Willems says the future isn’t about banning AI—it’s about assessing the judgment behind it. He favors reflective journals, prompt logs, design diaries. Spaces where the work isn’t just output—it’s thinking made visible. And SUSS’s Dr Wang Yue frames it differently: “AI frees us to focus on higher-order thinking.” Translation: Don’t fear it. Raise the bar. These are not professors trying to outwit the bots. They’re trying to rebuild learning around something AI can’t fake: discernment, originality, engagement.

Universities are good at rule-setting. Less good at meaning-making. They can restrict AI in exams. Set up detection tools. Require disclosures. All fine. But none of it fixes the core issue: students will keep trying to game the system if the system doesn’t feel like it sees them.

What students need—what we all need—is a clearer answer to the question: why does this matter? Why is this course worth our time? Why is this skill worth practicing manually? Why is this prompt something only we can answer? If we can’t answer that, then no policy will save us from drift.

This isn’t a tech story. It’s a cultural one. It’s not about plagiarism. It’s about presence. AI is not replacing student thinking. It’s revealing the absence of it—an absence that has been growing long before ChatGPT arrived. The students posting on Reddit? They’re not villains. They’re narrators of a quiet shift. A shift where education starts to feel like theater, and where AI isn’t breaking the rules—it’s exposing them.

We can’t fix this with stricter rubrics. We fix it by asking different questions. What kind of learning do we want to defend? What kind of attention are we willing to earn? And what kind of futures are we preparing students for—if the only thing we teach them is how to look like they know? Because the real danger isn’t that AI is doing too much.

It’s that we’ve forgotten how to care when it doesn’t.