During the 2026 National College Entrance Exam (June 7–8), six major AI platforms—Doubao, Alibaba’s Tongyi Qianwen, Tencent Yuanbao, Kimi, DeepSeek, and Baidu’s Wenxin Yiyan—will restrict or suspend certain features during exam hours, as required under the ‘Artificial Intelligence + Education’ Action Plan jointly issued by the Ministry of Education and four other government agencies. The specific restrictions vary per platform: Tongyi Qianwen, Kimi, and Wenxin Yiyan will disable image-based problem-solving and subject-specific answer generation; Tencent Yuanbao will suspend all image recognition functions and essay creation related to the college entrance exam; DeepSeek will halt image-based problem solving and analysis of science-related questions. Doubao’s customer service confirmed that its image-based Q&A feature will also be disabled, noting this has been standard practice over the years to uphold exam fairness; exact limitations will be detailed on each platform’s interface at the time. According to Red Star News on May 26, a Baidu representative stated no official notice regarding Wenxin Yiyan’s adjustments had been received yet, while iFlytek anticipated that ‘any large language model is likely to face similar restrictions.’
Notably, some netizens who tested the systems early on found that even after explicitly stating they were university students rather than high school examinees, the AI platforms still refused to provide answers to K–12-level academic questions, enforcing blanket restrictions. While this effectively deters cheating during the college entrance exam, it also inconveniences numerous university students currently in their final exam period—tasks such as drafting research papers or organizing scholarly references are blocked by the systems, flagged as potential cheating activities. This collateral impact has sparked considerable discussion online. Looking back at the 2025 exams, platforms like Tencent Yuanbao, Tongyi Qianwen, and Quark had already implemented comparable measures voluntarily; now, their inclusion in official governmental guidelines signals that such practices have officially become part of regulatory policy.
Red Star News (via The Paper) | Sina Finance
“高考期间AI工具将禁用”引热议，豆包、元宝、百度、科大讯飞回应 }