"Huanyu Zhiying ‘possesses the dual capability of simulating foreign‑media reporting and generating Chinese‑side response‑discourse text, and can introduce a specified guiding perspective to generate customised content,’" the school's announcement says — a compact description of a project that aims to turn foreign public opinion into a computational problem.
Huanyu Zhiying 1.0 and its claimed capabilities
Unveiled in April 2026 at Sun Yat‑sen University in Guangzhou, Huanyu Zhiying 1.0 (寰舆智应) is presented by its developers as an "LLM for global crisis response." The system — described by its creators as a research prototype — combines more than 10,000 simulated user profiles spanning a claimed 18 countries, and steps through five stages that include simulating foreign media, modelling audience reactions, and producing Chinese‑side response text. Its last step, according to the announcement, generates response copy intended to conduct "public‑opinion guidance" on different issues.
Training inputs: surveys, social media and fine‑tuned models
The synthetic population is built from multiple sources the developers cite explicitly. Sun Yat‑sen University’s own survey, The World Looks at China, polled respondents in eight countries — Australia, the United States, Britain, Canada, India, Malaysia, France and South Africa — using samples designed to reflect each country’s demographic composition. The developers say they combined those results with data from the World Values Survey and public social‑media data to construct simulated foreign users.
The system uses an AI model from Alibaba’s Qwen family, fine‑tuned on foreign coverage of China and Chinese media’s own international reporting. In practice the pipeline produces a simulated foreign‑media article — "think of a fabricated article written in the style of the BBC or Reuters" — has the synthetic population read and react to it, predicts changes in user sentiment and topic spread, and then uses those predictions to shape suggested responses.
This site is the portfolio.
OSINTSights runs on Cloudflare Workers, D1, R2, and Vectorize, with an AI pipeline on Hetzner ARM. Nubivance designed, built, and operates it. We do the same for clients.
See what we buildInstitutional sponsors and the research lineage at Sun Yat‑sen
Huanyu Zhiying was jointly developed by Sun Yat‑sen University’s School of Journalism and Communication — principally its AI and Computational Communication Laboratory — and the School of Software Engineering. The journalism school has an established research footprint: in 2017 it was approved to establish the Guangdong Provincial Key Laboratory of Public Opinion Big Data Analysis and Simulation, the first provincial‑level key laboratory awarded to a journalism school in China, according to the launch material.
Researchers work inside an explicitly state‑linked structure. The Guangdong Provincial Party Committee Propaganda Department and the China International Communications Group co‑founded and have jointly run research centres inside the journalism school since 2023 and 2019 respectively, the developers note.
Policy direction: from the Politburo to the 15th Five‑Year Plan
The project is linked in its own launch material to high‑level policy guidance. The developers cite a Politburo study session in May 2021 in which Xi Jinping called for "precision communication methods" tailored to "the audiences of different regions, different countries and different groups." The launch material says Huanyu Zhiying exists to implement the Central Committee’s recommendations for the 15th Five‑Year Plan, which call for stronger area and country studies and improved international communication effectiveness — a view echoed in a China Daily report that computing power, data and intelligent technology are "becoming important foundational elements of the international communication system."
Sun Yat‑sen’s dean, Zhong Zhijin, has publicly argued for targeted, differentiated communication grounded in cross‑national differences in media habits and attitudes, drawing on Geert Hofstede’s cultural dimensions as a conceptual starting point.
How technologists, policymakers, and foreign publics should read the system
- Technologists and modelers: Huanyu Zhiying uses survey‑based synthetic profiles, fine‑tuned LLMs and public social‑media data to model reactions. Researchers will note the project's reliance on mixed data sources (The World Looks at China, the World Values Survey, public social media) and on an Alibaba Qwen model; its technical claims remain those of its developers, who call it an "interim research result."
- Policymakers and international communicators: The developers explicitly tie the system to Central Committee directives and to the 15th Five‑Year Plan's communication objectives. The project exemplifies how state‑backed institutions can frame AI and data as tools to implement top‑level communication strategy.
- Foreign media and publics (including Australians): The system’s synthetic population drew on responses from real Australians in The World Looks at China survey, illustrating the concrete ways overseas survey data can be incorporated into simulation pipelines. The developers do not disclose which 18 countries are covered in the full synthetic population or which data source underpins which national profile.
Limits, intent and a pointed question
The creators themselves acknowledge technical limits: there is no public technical paper, benchmark or country‑level error rate, and the developers call Huanyu Zhiying an interim product that has so far "preliminarily shown potential." That modesty matters because existing research shows large language models can reproduce some patterns in human survey responses — a phenomenon researchers have called "silicon sampling" — but simulation is not forecasting. What is clear from the record is not that the system can predict how foreign publics will behave, but that state‑linked researchers are explicitly trying to treat foreign public opinion as a problem that can be modelled, stress‑tested and optimized before messages are issued into the real world.
If Huanyu Zhiying is a wind tunnel for narratives, the remaining question is operational: will the prototype stay in the lab as a research tool, or will it be moved into real‑world communications practice? The launch materials tie it to explicit policy goals; the public record so far does not say which path comes next.
