In an unprecedented move aimed at curbing cheating, China has temporarily shut down all AI chatbots and related tools across the country as millions of students sit for the annual national college entrance examination, known as the Gaokao. The nationwide suspension, which began on Monday, is expected to last until the completion of the exams later this week.
The move comes as concerns have mounted over the potential misuse of generative AI tools to generate answers during the high-stakes examination. While some argue the move is a necessary measure to maintain academic integrity, others have voiced concerns about the broader implications for AI development and access in China.
A report from Bloomberg confirmed the AI tool ban for all the local LLM-based models. Alibaba Group’s Qwen and ByteDance’s Doubao AI chatbots don’t return results for analyzing photos of a test paper during exam hours from June 7 to June 10. The Doubao bot even stated that an uploaded picture of a test paper was not in compliance with the local law and regulations.
Another screenshot, as shared by The Guardian, revealed DeepSeek telling its users that the service was not available in specific hours during the exams to ensure fairness in college entrance exams. DeepSeek is currently one of the leading generative AI tools in China that’s setting the global Gen-AI stage by storm with its low-cost and low-resource LLM models.
The Gaokao, often described as China’s toughest college entrance exam, is a crucial determinant of a Chinese student’s future academic and professional path. The Chinese government has a long history of implementing strict measures to prevent cheating, from signal blockers to facial recognition technology. The rise of sophisticated AI, however, presents a new and complex challenge.
Note that OpenAI’s ChatGPT is not officially available in China, although users can access the popular chatbot via VPN services.