A university professor is calling for stronger academic standards after discovering students used artificial intelligence to cheat on a recent exam [1].
The incident highlights a growing tension in North American higher education as institutions struggle to detect and regulate generative AI tools. While some professors have successfully identified misuse, others report that the majority of AI-assisted cheating remains undetected, threatening the validity of academic credentials.
The professor, writing in Nature, detailed catching one student in their own class using AI during an exam [1]. This discovery prompted a broader argument that universities lack the necessary policies and oversight mechanisms to prevent students from exploiting these tools to produce answers [1], [3].
Similar patterns are appearing across the continent. At Western University in London, Ontario, officials reported only three incidents of AI cheating since the start of the semester [2]. Despite these low official numbers, a Western University spokesperson said they suspect many more students are using AI tools covertly [2].
This gap between reported incidents and suspected usage is reflected in broader student data. A survey of post-secondary students in the U.S. found that approximately 70% admit to using AI for their assignments [3].
Educators are finding it increasingly difficult to distinguish between original thought and machine-produced text. One professor said in a CBC interview that they have seen a surge in AI-generated work and that it is becoming harder to tell what is original [2].
To combat this, the Nature author argues that the current approach to AI in the classroom is insufficient. "Without strong standards and smart oversight, artificial intelligence risks eroding the foundations of higher education," the professor said [1].
“"Without strong standards and smart oversight, artificial intelligence risks eroding the foundations of higher education."”
The discrepancy between the 70% of U.S. students admitting to AI use and the low number of caught offenders suggests a critical failure in current detection software and institutional policy. As generative AI becomes more sophisticated, universities may be forced to move away from traditional take-home or digital exams toward supervised, analog assessments to ensure academic integrity.



