Can AI Generated Work be Detected?: US Universities Question The Reliability of AI Detection Software
Written by: Munirah binti Mohamad Nizam | Credit: Anuj Chopra, The Star
Credit: The Star
6 OCTOBER 2026
WASHINGTON: US universities are moving away from the use of AI detection software as concerns rise over the reliability of these tools in detecting AI-generated content in students' assignments. The rise of AI chatbots has made cheating or better known among educators as “homework scams” effortless for students, allowing the credible solutions for written assignments to be generated just within seconds.
While the issue of “homework scams” is concerning, an increasing number of universities in the US, ranging from Yale to Cornell, have banned the use of AI detector software as evidence of academic cheating due to its unreliability. Studies have shown that AI text detectors often produce false positives, whereby authentic work has been mistakenly detected as AI-generated. Additionally, these tools have been found to be biased against non-native English speakers and are no longer effective in preventing cheating due to the emergence of "humanising tools” that make AI-generated work appear more like human work.
One highlighted example is Timothy Paustian, a professor from the bacteriology department at the University of Wisconsin-Madison, who devised a trap to identify AI-generated work. He inserted the hidden prompt into the assignment and managed to find out that sixty out of 350 students were caught using AI. However, this method is no longer effective, as current large language models (LLMs) can ignore such prompts, and students might have inspected the text before submitting it.
As AI cheating becomes more widespread, universities are rethinking the way to evaluate students. They have shifted away from using the detection tools towards developing new forms of assessment methods to assess students' understanding, which includes oral tests and video debate.
MIT reports warned that overreliance on AI tools can create the illusion of learning and lead to “cognitive surrender”, where students skip the critical thinking parts and rely entirely on AI tools. At the same time, some academics also believe that students should utilize AI as it could improve efficiency and prepare them for the modern workforce.
Comments