Chinese universities introduce regulations on AI use in academic writing

As AI models like DeepSeek advance, university students are increasingly using AI tools to assist in academic writing. However, an overreliance on AI and improper use are becoming more prevalent among Chinese students.

A survey by MyCOS, covering over 3,000 university faculty and students, found that nearly 60 percent use generative AI daily or multiple times a week. Among students, nearly 30 percent primarily use AI for writing papers or assignments, with some admitting to directly copying and pasting AI-generated content.

University professors and experts told China Media Group (CMG) that beyond using AI for generating text, some students have engaged in academic misconduct, such as falsifying research data, editing experimental images or replacing independent design work with AI-generated content.

In response, many Chinese universities have introduced regulations outlining the permitted and prohibited use of AI tools.

Fudan University recently issued guidelines banning AI use in research design, data collection, image creation, thesis writing and confidential content. Serious violations could result in failing grades and impact degree eligibility.

Tianjin University of Science and Technology last year also implemented AI-generated content detection in undergraduate theses, setting a maximum AI content threshold of 40 percent, a policy that will continue for 2025 graduates.

Meanwhile, multiple research teams in China are developing AI detection technologies to combat academic fraud. As AI technology evolves, anti-detection methods must also advance to maintain their effectiveness.

Beyond technology, experts emphasize the need for better academic evaluation and AI literacy.

Huang Yating, a researcher at Zhejiang University's School of Education, suggests that assessment methods should prioritize higher-order thinking skills.

"AI can replace repetitive and standardized tasks," Huang said. "Professors should design assignments that AI cannot easily replicate and encourage collaboration with AI rather than full reliance on it."