Xinhua shows off Xinhua Yudian agent at World Media Summit, betting on traceable fact-checking

Xinhua used the seventh World Media Summit in Beijing on November 7 to present Xinhua Yudian, its AI agent for news. According to an agency representative speaking at the summit, the platform draws on verified Xinhua content and helps users search for, check and use news information.

One detail is easy to miss: the platform was not launched on the day. It went live on 26 June 2026, and the summit was the occasion for a systematic presentation to international peers, along with an invitation to cooperate.

Summit materials describe Xinhua Yudian as an authoritative assistant for current affairs built on what the agency calls a clean and reliable corpus. The product has four modules: AI chat, subscriptions and notifications, a knowledge base, and an assistant hub.

Seven functions sit inside those modules: studying theory, searching primary sources, receiving news recommendations, asking questions about the news, verifying information, analysing topics and preparing articles. Xinhua frames the goal as moving past the usual distribution of mainstream media content towards deeper knowledge and services.

On technology, the agency says it independently develops the core architecture, fundamental algorithms and main modules. To suppress hallucination and improve traceability, the system combines multi-channel knowledge-base retrieval, knowledge graphs and deep semantic search, backed by a three-tier protection system covering the knowledge infrastructure and the whole AI application process.

On scale, Xinhua points to 95 years of reporting. For major statements by Chinese President Xi Jinping alone, it says the system holds more than 100 volumes of documents and hundreds of thousands of full-length reports, supported by purpose-built knowledge graphs and entity-relationship networks to keep retrieval and citation accurate.

Deployments so far include public-opinion research, policy studies and help with drafting. Xinhua also invited foreign media organisations to cooperate, offering to share experience in building specialised text corpora and to discuss how historical reporting and professional knowledge should be organised, while seeking partners for multilingual and cross-cultural services.

The context of the summit matters as much as the demo. Earlier in the programme Xinhua floated a global initiative to develop standards for AI use in media, and participants adopted a Beijing Consensus. Presenting the agent after the standards push sends a clear signal: credibility and traceability are being treated as the entry ticket for media AI.