Students Are Humanizing Their Writing—By Putting It Through AI

Students don’t want to be accused of cheating, so they’re using artificial intelligence to make sure their school essays sound human.

Teachers use AI-detection software to identify AI-generated work. Students, in turn, are pre-emptively running their original writing through the same tools, to see if anything might be flagged for sounding too robotic.

Miles Pulvers, a 21-year-old student at Northeastern University in Boston, says he never uses AI to write essays, but he runs all of them through an AI detector before submitting them.

“I take great pride in my writing,” says Pulvers. “Before AI, I had peace of mind that whatever I would submit would be accepted. Now I see some of my writing being flagged as possibly being AI-generated when it’s not. It’s kind of annoying, but it’s part of the deal in 2025.”

AI detectors might sound the alarm if writing contains too many adjectives, long sentences and em dashes—one of my own favorite forms of punctuation. When that happens to Pulvers, he rewrites the sentences or paragraphs in question. He tests the essay again, as often as needed until the detector says it has a low probability of bot involvement.

He and other students I interviewed say teachers in recent months have become more suspicious of students. That’s often for good reason: Many do   use AI to cheat .

“They’re using AI to write their essays, and then they use AI detectors to humanize them,” says Leticia Villaseñor, a history teacher at a private Los Angeles high school, where she regularly uses an AI detector to catch cheaters. “I try to have my students do as many written assignments as possible in class, because if it’s homework, half of them will use AI.”

I’m not a robot

For students who swear they’re not having a chatbot write their essays, trying to prove their humanity has become a challenge.

Marcus Wooler, one of Villaseñor’s students, said some students are wrongly accused. A friend of his had to show a teacher his Google Doc history to clear himself. Wooler decided he should run his honestly crafted essays through an AI detector before submitting them. “It’s a good precaution,” says the 16-year-old.

Sometimes, it’s the students who use grammar-checking tools—or who learned to write more formally—who get dinged, says Devan Leos, co-founder of Undetectable AI, which makes a popular AI detector and “humanizer.” Some red flags include the use of unique words such as “delve” and “tapestry” and phrases like “valuable insight” and “crucial role.”

“It’s like you get penalized for being a proper writer,” Leos says.

Harrison Checketts, a 16-year-old student who takes Villaseñor’s history class, says he once ran an English paper he wrote himself through five different AI detectors and they all showed different results. Some were highly confident that parts were written by AI; others said there was no chance a machine did it. He decided not to use AI detectors after that. “The teachers are the real AI detectors,” he says. “They can tell.”

Claire Krieger, a 20-year-old student at Fordham University in New York, runs all of her papers through AI detectors but says they can provide a false sense of security.

One of the tools she used for an English paper last year flagged some quotes she cited from a book, but that’s it. Her professor used a different AI detector, which flagged a sentence Krieger says she wrote. She met with her professor in person, explained that it wasn’t AI-generated and affirmed her commitment to academic integrity.

A ‘police state’

The stakes for students accused of submitting AI-generated work are high. A University of North Georgia student was placed on academic probation last year after being accused of plagiarism. She told numerous media outlets that she used Grammarly, an AI writing assistant, to correct spelling and grammar mistakes—not to cheat. She also partnered with Grammarly to produce educational videos about how to use the tool for schoolwork.

The incident prompted Grammarly to create a new authorship tool. It tracks the writing process, showing where text is typed into a document or pasted, as well as which parts of a document are created or modified with AI. When the paper is complete, a report is generated, which students can show teachers if there is any question about the source of their work.

Jenny Maxwell, Grammarly’s head of education, says the fear around AI use in schools is causing students to spend a lot of extra time rewriting to sound more human. More than 500,000 people use Grammarly’s AI and plagiarism detection tool each week, and the majority are students, a spokeswoman says.

These days, Maxwell says, it’s a “police state of writing.”

Write to Julie Jargon at Julie.Jargon@wsj.com

Editor’s note: We ran this edited column through an AI detector that determined it was 99% likely to have been written by a human. We assure you, the following is 100% human.