How to Fix the Cheating Crisis

The problem isn't the people or the tools; it's a failure of deterrence.

Courtesy Nguyễn Hiệp/Unsplash.

Everyone, it seems, is worried about academic dishonesty. Professors and administrators are pulling their hair out over the ways artificial intelligence (AI) lets students cheat on take-home assignments—and even helps them apologize when they’re caught. Over on X, meanwhile, commentator and provocateur Helen Andrews launched a firestorm by arguing—as she has previously—that rising Asian immigration has already contributed to a cheating crisis.

These concerns are not unmerited. Data from 2024, now outdated, suggest that as much as 10 percent of student work involved at least some amount of AI use. AI-enabled cheating has nearly quadrupled by another measure. And while we shouldn’t make judgements about individuals, Andrews is correct to note the startling prevalence of cheating among East Asian students: in a survey of students at four Chinese public universities, for instance, more than half admit to academic dishonesty by some measures (although that’s not that different from American rates).

While the causes of rising cheating are interesting, fixating on technological or social determinants of the issue distracts us from asking how we can actually solve the problem. When we pivot to that question, it becomes obvious that the cheating crisis is not the result of forces beyond our control, but of our unwillingness to exercise control in the first place.

Why do students cheat? It’s the subject of countless think pieces, which blame everything from subcultural norms (a la Andrews) and insufficient student motivation (“class is too boring!”) to an amorphous cheating “culture.” Such analyses look a lot like classic “root causes” theories of crime, trying to discern the fundamental variable behind anti-social behavior. And, like those theories, they end up demanding profound social change to address that behavior, whether that be an uprooting of pedagogical practices or shutting off the flow of high-skilled immigration.

But, as I argued in last fall’s issue of City Journal, this mode of analysis was never very helpful for actually reducing crime. Indeed, it left criminologists insisting that we could not bring down the sky-high crime rates of the 1970s and 1980s without first resolving all the evils of society—capitalism, racism, and so forth. Yet somehow, crime plummeted in the 1990s, even as the problems of poverty, inequality, and racism either remained constant or got worse. What changed?

The answer is, in large part, that law enforcement across the nation embraced deterrence—the art and science of discouraging people from engaging in bad behavior by threatening a cost worth more than the benefit obtained from engaging in that behavior. In the context of crime, deterrence doesn’t change people’s tendency to commit crime—it doesn’t make them richer or less subject to social forces. Instead, it raises the cost of committing a crime in spite of whatever is inducing them to do so.

In his classic formulation, economist Gary Becker argued that deterrence is a function of the swiftness, certainty, and severity of punishment. Given that formula, it’s easy to see why cheating has grown so rampant on college campuses: punishment is neither swift, nor certain, nor severe.

Actual identification of cheating is rare. Fewer than 10 percent of cases of academic dishonesty are reported or formally sanctioned, according to one review. Swiftness is hamstrung by the legal environment: long-standing precedent allocates students strong due process and free speech rights, and lawyers can make a lot of money suing deep-pocketed schools.

When offenders are identified, punishments are rarely severe. At the University of Ohio, for example, most receive probation or a formal reprimand. Expulsion, by contrast, is a little-used tool in modern higher education. Take the recent student encampments on many campuses, which often led to actual criminal acts: “Students were suspended and expelled in the ’60s and the ’80s, but more recently we’ve seen universities be much more lenient with student protesters,” Thai Jones, a Columbia social-movement historian, told the Washington Post.

Why are universities so hesitant to punish students? The reluctance is partly pecuniary—each student kicked out means one fewer tuition payment. That’s particularly so with foreign students, who are more likely to pay full freight. But with admission rates still near rock-bottom lows, each expelled student can easily be replaced by a willing former reject.

The legal barriers also discourage enforcement. Gabriel Rossman, a professor of sociology at UCLA, told me that universities’ burdensome process for establishing student innocence creates prohibitive backlogs, often causing cheating cases to languish for months.

Many professors just don’t want to bother with the hassle. “I hear from people who don’t file reports and it’s like the cop character in a movie who is like, ‘I just don’t give a shit, 10 years to my pension,’” Rossman said. That laziness can often blend with an unwillingness to be confrontational—as at Columbia, where two janitors claim they were told students spray painting swastikas on school property were “exercising their First Amendment rights.”

A growing backlog of cheating cases, though, is actually a consequence of inaction. If students were more afraid of breaking the rules, they wouldn’t generate so many cases. Less punishment now, paradoxically, can generate the need for more punishment later.

The solution, then, is to re-establish deterrence by mounting an aggressive campaign of rule enforcement. Top universities need to think about expelling not just one or two students, but ten, 50, or even 100. That will send a clear signal to their peers that they can no longer get away with rule breaking.

Expulsion may sound severe. But increasing the severity of punishment is much easier than increasing its swiftness or certainty. And, because most university students are rational actors—or, at least, more rational than most criminals—they should respond to an increased expected future cost of misconduct by decreasing that behavior.

In the short run, this approach will mean a burst of resources dedicated to adjudicating these cases—the best way to cut through burdensome process is to add more judges. But swiftly and aggressively prosecuting every identified violation will soon reduce the flow of new rule-breakers to a trickle. At that point, additional enforcement will be relatively easy—because deterrence will be re-established.

Cultural norms—around AI use, or imported from other nations—are not amorphous and unchanging laws of nature. They are behaviors that respond to incentives in both the short and long run. If Andrews is correct that East Asians cheat more often than Americans, or if AI cheating is widespread, then the best solution is not to stop allowing high-skilled immigration or capitulate student whining. It’s to aggressively enforce our rules against cheating, assimilating cheaters to our norms through enforcement.

In this regard, the cheating crisis is like many other crises. It is rarely worthwhile to ask why, at root, bad behavior happens. It is almost always better to ask what we can do to stop it—that’s the tractable question.