Twelve point nine million students sat China’s national college entrance examination, the gaokao, on June 7 this year — one of the largest standardised tests on the planet, and one now being deliberately reshaped for an AI economy. Thirty-eight new undergraduate majors have been added to the national catalogue for the 2026 admissions cycle, including Embodied Intelligence and Brain-Computer Science and Technology, reflecting emerging industries and technological innovation that Chinese universities are now explicitly recruiting toward. At the same time, and for the first time, Beijing is exporting a version of its meritocratic testing philosophy to international applicants through a new compulsory entrance exam — a signal that China’s higher-education strategy in 2026 is as much about quality control over who enters its universities, foreign or domestic, as it is about domestic reform.
The dual moves — AI-aligned domestic reform and a new international admissions gate — mark the most significant shift in Chinese higher-education access policy in years, and they are unfolding against explicit government warnings about AI’s growing, and sometimes fraudulent, role in exam preparation itself.
Context: an exam reforming itself to resist the technology reshaping everything else
The gaokao’s core tension in 2026 is almost paradoxical: China is simultaneously building AI into every layer of its education system through a national AI literacy action plan, while working hard to ensure the exam itself remains resistant to AI’s most obvious shortcut — predicting or generating answers to test questions. The Ministry of Education issued a formal “2026 Gaokao Warning Information” notice ahead of the exam, urging students and parents to beware of false advertising from self-styled experts and platforms claiming that AI could predict composition topics. The warning proved prescient: journalists who prompted five mainstream AI models — including ChatGPT, Claude, Gemini and China’s own Kimi and Douyin — to predict the 2026 essay topic found that not a single model’s guess matched the actual question, which authorities say was deliberately designed with a “clear anti-prediction and anti-formula orientation.”
That anti-prediction posture reflects a broader recalibration of what the exam is meant to measure. Officials have stated plainly that the gaokao increasingly focuses on critical thinking and problem-solving rather than memorisation and rote test-taking tricks — a shift that runs directly counter to a decades-old tutoring-and-prediction industry built around gaming exactly those tricks.
AI’s real role in the 2026 gaokao: assistance, not authorship
Even as officials warn against AI-generated exam predictions, artificial intelligence has become deeply embedded in the gaokao’s supporting infrastructure. According to Chinese state media, more than 10 million users engaged Baidu’s AI application assistant on a single day after the 2025 gaokao results were released, and major platforms including Quark, Baidu and Douyin have built AI-powered university-and-major recommendation tools that let students input scores, provincial rankings and subject combinations to receive tailored application guidance — a service that previously cost families thousands, sometimes tens of thousands, of yuan through private consultants. Security technology has advanced in parallel: AI-assisted invigilation systems and biometric identity verification are now standard at exam sites, and Guangdong province’s Yangjiang city has deployed a “2+1” security-screening system combining intelligent security gates with metal-detector checks at all fifteen local exam sites.
The official framing, repeated across state commentary, is unambiguous: AI can organise information and support logistics, but “it should not be treated as an oracle” — a formulation that draws a careful line between AI as an administrative and revision tool versus AI as a substitute for judgment, fairness or diligence.
A new gate for the world: the China Scholastic Competency Assessment
Perhaps the more structurally significant change for global education is one aimed entirely at foreigners. Beginning with the 2026 intake, most international applicants seeking undergraduate study in China will be required to sit the China Scholastic Competency Assessment (CSCA), a centrally designed test intended to benchmark students from different education systems against a common academic standard. The exam becomes compulsory this year for recipients of Chinese government scholarships and will phase in more broadly, becoming mandatory for all international undergraduate applicants by 2028. It will be delivered primarily as a remotely proctored online test, with offline centres available in some countries.
Richard Coward, CEO of the admissions agency Global Admissions, has called the policy “one of the biggest changes” he has seen affecting international students applying to study in China, describing it as part of a broader global pivot — echoed in Canada’s and the UK’s own visa reforms this year — away from enrolment quantity and toward quality and standardisation. Scholar Gary Postiglione has noted a structural parallel to the domestic gaokao itself: China’s centralised, meritocratic testing culture leaves little room for informal admissions pathways, meaning the CSCA is likely to function as a genuine academic filter rather than a formality — though he cautions that students who have already studied in Chinese, or who come from mathematics-strong education systems, may find the standard easier to clear than others, raising early questions about equity of access across different national school systems feeding into China’s international student pipeline.
Why this reshapes the calculus for students and universities alike
For Chinese students, the 38 new majors and the exam’s harder pivot toward interdisciplinary, application-based questions signal where Beijing wants its next generation of graduates concentrated: AI, brain-computer interfaces, and fields tied to the 15th Five-Year Plan’s ambition of a leading global position in AI industry applications. Universities gain more latitude to align enrolment with these priorities, and — per the Beijing Education Examination Authority’s detailed 2026 schedule — elite institutions such as Tsinghua and Peking University continue reserving substantial local quotas even as national reform pushes toward more holistic, application-oriented evaluation across 26 provinces now using variants of the “3+3” subject-choice model.
For the growing population of international applicants — currently drawn heavily from Central Asia, China’s regional partners and increasingly from students seeking alternatives to a tightening US and UK visa environment — the CSCA represents a new, non-trivial hurdle layered on top of language requirements and existing university-specific admissions criteria. It also positions China’s higher-education system as an increasingly rules-based, standardised alternative to the more discretionary and increasingly restrictive Big Four systems, even as it introduces its own new form of selectivity.
The unresolved question, flagged by admissions specialists themselves, is whether a single centrally designed assessment can fairly benchmark applicants coming from wildly different national curricula — or whether, like the gaokao’s own long-debated regional quota disparities, the CSCA will simply relocate old fairness arguments onto a new, internationalised stage.
