What happened when students tried using AI during China's college entrance exam

Companies like Alibaba, Tencent, ByteDance, and DeepSeek have suspended photo-recognition and question-answering functions in their AI chatbots to prevent cheating during this critical testing period, which determines university placements for millions.

This year, students hoping to leverage AI tools were met with restrictions, with screenshots shared on Weibo, China’s popular social media platform, capturing one user’s attempt to upload an exam question to Doubao. The app responded, “During the college entrance examination, according to relevant requirements, the question answering service will be suspended.” Even when the user insisted, “This is not the college entrance examination,” the response remained unchanged. Similarly, DeepSeek, a generative AI tool that surged in popularity in 2025, blocked services during exam hours, citing the need “to ensure fairness in the college entrance examination,” according to Bloomberg.

Other platforms, including Tencent’s Yuanbao and Moonshot AI’s Kimi, also disabled photo-recognition features, which could otherwise analyse test papers. These measures have frustrated some university students who rely on these tools for assignments

The AI blackout is part of a broader anti-cheating arsenal whereby Chinese authorities have deployed AI-powered monitoring systems in provinces like Jiangxi to detect “abnormal behaviours” such as whispering or suspicious glances, with footage reviewed post-exam for violations, per the Global Times. Stricter entry protocols include biometric identification, enhanced device screening, and radio signal blockers to neutralise illicit communication devices.

The voluntary suspension of AI features by tech giants, without formal public announcements, suggests a coordinated effort to align with regulatory expectations and uphold exam integrity.