Disposable Aprons Proposed to Block Miniature Cameras
Special Exam Paper Technology That AI Cannot Read Also Discussed
"Add QR Codes to Each Exam Paper to Trace Leaks"

With two days to go before the 2027 College Scholastic Ability Test (CSAT), concerns about cheating with artificial intelligence (AI) are growing, and a variety of ideas to prevent it are emerging. Suggestions range from making test takers wear disposable aprons or inserting QR codes into exam papers to printing special patterns that prevent AI from reading the questions.

An image illustrating a proposal to cover test takers' clothing buttons with tape and make them wear disposable aprons to prevent cheating with AI devices. ChatGPT

An image illustrating a proposal to cover test takers' clothing buttons with tape and make them wear disposable aprons to prevent cheating with AI devices. ChatGPT

View original image

"Make Them Wear Aprons and Add QR Codes"... A Flood of Ideas to Prevent Cheating

According to education officials and other sources on November 17, a range of measures to prevent cheating with AI devices during the CSAT on November 19 have recently been proposed on online communities and the e-People petition portal.


On November 16, a post titled "Disposable Aprons Should Be Required for the CSAT" appeared on the online community DVD Prime. The author proposed providing test takers with nonwoven-fabric aprons to prevent them from photographing exam papers with micro-cameras hidden in their clothing. Similar proposals were made on other online communities, along with suggestions to cover cameras attached to clothing with white masking tape.

A response to "AI cheating" posted on an online community on the 16th. Screenshot from DVD Prime.

A response to "AI cheating" posted on an online community on the 16th. Screenshot from DVD Prime.

View original image

Another idea was to add unique QR codes and watermarks to exam papers. On November 13, a proposal was submitted to e-People calling for unique identification numbers to be printed on every page of the exam paper or in the margins around each question. The idea is that if each exam paper is assigned a different number and linked to a test taker's registration number, the test taker who received a paper later leaked outside the test center could be traced through a photo of it.


Other suggestions included using AI monitoring systems to detect unusual behavior by test takers, requiring those with hair of a certain length to tie it back, and restricting clothing with buttons or collars.

"Readable by People, Not by AI"... Special Exam Papers Also Suggested

Another proposal was to change the exam papers themselves so that AI could not read the questions. The idea is to print special patterns on the text so people can read it normally, while AI would have difficulty recognizing images captured by a camera.

An image illustrating a proposal to print tiny dots and noise patterns in the margins of exam papers or around text to interfere with AI character recognition. ChatGPT

An image illustrating a proposal to print tiny dots and noise patterns in the margins of exam papers or around text to interfere with AI character recognition. ChatGPT

View original image

In 2018, researchers at Cornell University proposed a technique that makes subtle changes to images of text, causing optical character recognition (OCR) systems to misidentify them as different words. In 2023, researchers at Ningbo University in China also presented a technique that interferes with text recognition by manipulating image backgrounds.


However, these technologies have not yet been commercialized to the point where they can be used at exam sites. Their effectiveness may vary depending on the camera angle or lighting, and if the text is altered enough to interfere with AI recognition, test takers may also have difficulty reading the questions. Another limitation is that the latest AI can recognize blurry or altered text.


Aprons may conceal cameras attached to clothing, but they are unlikely to block smart glasses. QR codes and watermarks are also intended for tracing after the fact, rather than catching cheaters on site. As a result, security industry experts say that, rather than relying on technological measures, a more practical solution is to prevent smart devices from being brought into test centers and strengthen proctoring.

"I Can't Sleep"... Government Steps Up Its Response to Test Takers' Anxiety

Complaints related to AI misconduct filed on the 17th. Screenshot from e-People.

Complaints related to AI misconduct filed on the 17th. Screenshot from e-People.

View original image

As controversy over AI-assisted cheating spreads, test takers are becoming increasingly anxious. More than 2,400 posts were submitted to the questions-and-answers board on the Korea Institute for Curriculum and Evaluation's CSAT website between November 15, when news broke that a test taker had confessed to cheating on the tax accountant exam, and November 17.


One author, who said they were taking the exam for the fourth time, pleaded, "Please, do something. I can't sleep," adding, "Please don't dismiss this as just an emotional appeal from young children." Another test taker said stricter enforcement could mean even the earplugs they normally use might be restricted, and called for clear rules on their use.



On November 17, the government announced measures to prevent cheating with AI devices. It plans to cover the buttons on all test takers' clothing with stickers and check their sleeves and collars, while inspecting for smart glasses and sending emergency text alerts to identify mobile phones that have not been turned in. Even after the exam ends, authorities will work with telecommunications companies and global technology companies to track suspected cases of cheating.


This content was produced with the assistance of AI translation services.

© The Asia Business Daily. All rights reserved. Unauthorized AI training and use prohibited.

Today’s Briefing