evidoria

← Back to browse

Good practice Imported

Guarding the Gaokao — AI Exam-Hall Patrols and an Industry-Wide Chatbot Freeze

China · Guangzhou · See the China profile

Top 73% 40/100 · Ask Evidence Copilot about this practice

Guangdong, Hainan and Shandong deployed AI 'intelligent patrol' cameras across Gaokao exam halls, while Alibaba, Tencent and Moonshot AI froze chatbot photo-recognition during China's 13.4-million-student exam to stop students photographing papers for answers.

Guarding the Gaokao — AI Exam-Hall Patrols and an Industry-Wide Chatbot Freeze

Details

Promoter
Guangdong Provincial Department of Education; Alibaba (Qwen), Tencent (Yuanbao), Moonshot AI (Kimi)
Period
2024–2025 (annual June exam window)
Keywords
K-12 assessment, exam security, AI industry governance

Description

China's national college-entrance exam, the Gaokao, draws a huge cohort each year — a record 13.42 million students registered in 2024. To guard against cheating, the provinces of Guangdong, Hainan and Shandong adopted 'artificial intelligent patrol systems': camera-based monitoring in examination and confidential rooms that China's Ministry of Education described as part of an effort to 'actively promote real-time intelligent inspection of examination rooms.'
The integrity effort extends beyond government surveillance to the AI industry itself. During the 2025 exam window (June 7–10), Alibaba's Qwen and Tencent's Yuanbao chatbots disabled photo-recognition features nationwide, responding to camera-based homework queries with a message that the function was 'temporarily frozen during exam hours.' Moonshot AI's Kimi did the same, while ByteDance's Doubao kept photo recognition active but flagged exam-paper images as non-compliant.
The platforms cited exam fairness directly: 'To ensure the fairness of the college entrance examinations, this function cannot be used during the test period.' The dual approach — physical AI surveillance in exam halls plus voluntary, coordinated feature freezes by the country's largest consumer AI apps — reflects a governance model in which state education authorities and private AI vendors act in tandem around a single high-stakes national event.
No independent effectiveness data (e.g., change in confirmed cheating incidents) has been published for either measure, and camera-based 'intelligent patrol' surveillance raises its own student-privacy questions that have not been publicly addressed in the available reporting.

Read the full analysis: https://fortune.com/asia/2025/06/09/alibaba-tencent-freeze-ai-tools-gaokao-china-exam

Implementation

Implementation detail (cost, timeline, staffing, conditions for success) is not yet available for this practice.

Do you run this practice? Claim it — verified implementers get a public contact pathway and can propose corrections.

Data sources

Where this practice's information was retrieved from, and when.

Attachments

Similar practices you may find useful