AI tools disabled across China amid college entrance exams

Held annually from June 7 to 10, China’s Gaokao exam involves 13.4 million students vying for spots at 147 prestigious universities, with strict anti-cheating measures such as drones, device bans, and electronic jammers safeguarding the process

Leading Chinese artificial intelligence companies temporarily disabled certain functions of their AI chatbots during this year’s Gaokao examinations, the country’s most crucial academic test. The move, aimed at curbing the misuse of AI tools for cheating, came amid rising concerns over the potential of generative AI to compromise exam integrity.

Held from June 7 to 10, the Gaokao is a high-stakes college entrance examination taken by more than 13.4 million students across China. The results determine eligibility for limited seats at leading institutions, including 147 top-ranked universities under the government’s “Double First-Class” initiative. Given its significance in shaping students’ academic and professional futures, the exam is conducted under strict surveillance, with measures such as electronic jammers, banned devices, and even drones in place to prevent malpractice.

AI tools curbed for fairness

This year, however, concerns extended to AI-powered tools. In response, major tech firms including Alibaba (Qwen), Tencent (Yuanbao), ByteDance (Doubao), and Moonshot AI (Kimi) disabled image recognition and question-answering features in their chatbots during the exam period. These tools, which are capable of solving academic problems and processing visual data, are seen as potential risks if accessed by students during exams.

According to reports, when users tried to upload images or ask academic questions, platforms like Yuanbao and Kimi displayed automated messages noting that “such features have been suspended to ensure fairness in the college entrance examinations.” ByteDance’s Doubao permitted image uploads but declined to respond to test-related content, citing regulatory compliance. Similarly, Alibaba’s Qwen reportedly avoided analyzing any content resembling exam questions during testing hours.

China sets AI education guidelines

In a broader context, China’s Ministry of Education has recently introduced a national framework for AI use in education, specifically targeting the integration of generative AI in schools. The guidelines advise against students independently using AI tools that generate open-ended content and recommend that educators use such tools to enhance—not replace—traditional teaching methods. Middle schoolers are encouraged to learn about AI-generated content structures, while high school students can delve deeper into the underlying technologies.

A directive published in China Daily by the government’s education committee for 2025 underscores the importance of ethical, age-appropriate AI use, and warns against academic dishonesty, over-reliance on automation, and privacy violations. Together, these actions signal China’s commitment to upholding the integrity of its academic system while cautiously embracing AI in the classroom.