Does Traditional Plagiarism Still Matter?

In the age of generative AI, the idea of copy‑and‑paste plagiarism sounds almost quaint.

Do you have an upcoming essay that you can’t be bothered to write? Just use ChatGPT or Claude to generate it for you. Don’t want to write that article? Just feed the topic to Gemini or Grok.

Not only is AI generated content easier to produce, it also carries fewer risks if you are caught. The uncertainty around AI detection means that instructors have a more difficult time proving AI cheating than “traditional” plagiarism. 

In a time where it is widely believed that AI articles either have or will soon surpass the number of human articles, it can seem like copy-and-paste plagiarism is a thing of the past.

However, the data says otherwise.

Buried in the footnotes of a report, the academic integrity company Turnitin showed that traditional plagiarism hasn’t gone anywhere. Though AI has definitely become the focus when it comes to academic integrity, copy-and-paste plagiarism is still a serious problem and one that isn’t going away.

But that raises one major question: Why?

Understanding the Data

Beginning this year, Turnitin has been publishing a quarterly “Learning Integrity Insights” report. It contains data gleaned from the company’s massive presence in the education space, which includes more than 16,000 institutions all over the world.

In their first report, Turnitin focused heavily on how teachers and administrators alike were pushing for greater transparency over the use of AI, rather than outright banning it. However, this is being frustrated by a “policy gap” where less than half of schools have an AI policy in place.

However, an aside to the report analyzes the amount of “traditional” plagiarism that is still taking place. The report analyzes two three‑year periods: the first from November 2019 to November 2022, the three years before the launch of ChatGPT, and April 2023 to April 2026, the three most recent years available.

During the pre‑ChatGPT period, Turnitin found that 6.86 % of submissions contained more than 80 % similarity. From the more recent period, that number only dropped to 6.16 %. While that does represent a 10 % drop, it is nowhere near the precipitous decline that many would have expected.

The two datasets were also similar in what was plagiarized. In the pre‑ChatGPT dataset, 88.25 % of the papers matched other student works. In the more recent dataset, that number increased to 91 %. In the pre‑ChatGPT set, some 10.75 % of papers matched internet sources; meanwhile, the more recent dataset saw the number drop to 7.8 %.

What’s striking about these numbers is how little they changed. While there was some drop‑off in “traditional” plagiarism, the drop was not very dramatic and may not even be statistically significant.

That’s because this report only looks at papers with 80 % or greater similarity. Much of the copy‑and‑paste plagiarism that takes place focuses on paragraphs and passages rather than whole‑cloth documents.

That said, the comparison is the best apples‑to‑apples comparison with AI generation. Students who want to completely avoid writing an assignment have had AI to choose from and, despite that, roughly the same percentage are turning to copy‑and‑paste.

Why would they do that?

Why Copy and Paste?

The Turnitin report doesn’t go into detail about why students are still engaging in “traditional” plagiarism. As such, I am forced to speculate.

What I think is most telling is that, both pre‑ChatGPT and post‑ChatGPT, the majority of copied content came from their student library. Contract cheating is an incredibly complicated and nuanced problem.

For every essay‑mill paper purchased, many times that number is obtained through less formal means. Whether it’s getting a paper from a friend who took the class before or having another student do your work for you. This practice predates not just AI, but the internet itself.

Most students understand that copying and pasting from the internet is incredibly risky. Tools like Turnitin have made such plagiarism difficult to get away with. The dark ages of academic plagiarism were largely over by the late 2000s. 

So, for the past 15 to 20 years, copy and paste plagiarism from the internet has been incredibly risky, but still surprisingly common. Still, this is a big part of why contract cheating became such a hot topic in the 2010s and early 2020s. Students saw it as a way to circumvent plagiarism detection tools.

However, one would think that AI would have been the nail in the coffin for this practice. But AI has produced its own set of problems. Though students have always been scared of being falsely accused of plagiarism, AI has ramped those fears up to a new level. 

The reason is simple: there’s still a great deal of uncertainty as to what will and will not be detected as AI‑generated. AI detection tools are a black box and false positives are a real concern. 

Much of this concern is from legitimate students. However, it also impacts those who are trying to use unethical shortcuts. AI is seen as risky and uncertain, so purchasing a paper or using one from a previous year (especially if it predates AI) may be seen as a safer option. 

However, it isn’t. The old plagiarism detection tools didn’t disappear the day ChatGPT was launched. They are still around and in widespread use. Still, it’s easy to see why students may continue using the “older methods.” We saw something similar with students turning to essay mills, as internet-based plagiarism became more risky. 

As a result, traditional plagiarism and contract cheating will continue to be a serious problem for academia and it may even get worse if AI detection tools improve. 

Bottom Line

The headline news here (and the bad news overall) is that the rise of generative AI has not significantly reduced the amount of “traditional” plagiarism in academia. Copy and paste plagiarism still exists and in largely equal numbers.

This means that AI has added a significant burden to academia rather than simply shifting the focus. The challenges around AI are bad enough without other forms of plagiarism remaining steady.

However, also buried in that same report is that less than half of all schools have an AI policy. This is unconscionable. Even if schools ignored the warning signs, it has been nearly four years since ChatGPT was released. There has been more than enough time to draft and implement an AI policy.

Leaving Instructors to figure this out on their own is unfair to both them and their students. That, in turn, is likely part of why traditional plagiarism is still so prevalent. With so much uncertainty around AI, is it any wonder that students would be scared of it, even when they are trying to cheat?

Solutions to academic integrity issues begin with good policies that are well-implemented. Without that, both AI and traditional plagiarism will continue to be a major problem for all involved.