No foolproof way to stop students using AI to cheat in remote exams, expert says

A leading computer science professor has stated there is no realistic way for universities to fully prevent students from using AI tools to cheat in online exams, according to a report by Finnish public broadcaster Yle.

Teemu Roos, a professor of computer science at the University of Helsinki, told Yle that the only theoretically airtight solution—installing 360-degree cameras in students’ homes—was impractical. Even blocking access to certain websites on exam devices would fail, he noted, as students could simply use AI on another device.

“Completely preventing the use of prohibited aids seems extremely challenging,” Roos said.

The comments follow a Yle investigation revealing that health care courses at Finnish universities of applied sciences could be passed in hours by using AI to answer questions and perform drug calculations. Roos acknowledged that while in-person, pen-and-paper exams would be the simplest safeguard, many instructors have already shifted back toward supervised on-campus testing where possible.

He stopped short of advocating for the abandonment of remote exams, suggesting instead that some subjects could replace multiple-choice tests with assignments requiring reflection on the learning process—tasks he described as harder to outsource to AI.

Roos added that cheating on minor courses ultimately harms students themselves more than the institution, as it undermines their own learning without significantly affecting their degree.