Six major AI platforms restrict some features/services during the 2026 Gaokao
What happened
It has been reported that six major Chinese AI platforms — including Baidu (百度), Alibaba (阿里巴巴), Tencent (腾讯), iFLYTEK (科大讯飞), Huawei (华为) and NetEase (网易) — limited certain features and services during the 2026 Gaokao. Reportedly, restrictions covered functionalities most likely to be used for cheating: real‑time document upload, advanced image recognition, code execution and some long‑form question‑answering tools. Companies said the moves were taken to reduce the risk of exam‑related misuse and to comply with guidance from education authorities.
The step is familiar. Chinese platforms have previously throttled or disabled features around major national exams after regulators and schools warned that powerful language and multimodal models could be misused to generate answers or interpret test materials. This year’s action appears broader in scope and, reportedly, involved coordinated timing among the industry’s largest general‑purpose models and associated toolkits.
Why it matters
For Western readers: the Gaokao is China’s high‑stakes national college entrance examination — a two‑ or three‑day test that heavily influences university placement and life chances for millions of students. When AI tools can generate essay responses, parse images of questions or execute code to solve problems, regulators and universities see an acute integrity risk. Can innovation coexist with a system where a single exam can determine futures?
There is also a regulatory and geopolitical backdrop. Beijing has been tightening oversight of AI, data flows and online education; companies are incentivized to err on the side of restriction to avoid sanctions or reputational fallout. Reportedly, the temporary curbs will be lifted after the exam window, but the episode raises broader questions about how platforms balance product capability, compliance and social responsibility in a market where government guidance and public trust carry heavy weight.
