UN turns to Google to make its global data ready for AI agents

The United Nations is teaming up with Google to make its statistics easier for AI systems to reach, read, and reuse. The new setup, called the UN System Data Commons, sits on Google’s open source Data Commons platform and lets users search across UN agencies in plain language instead of poking around a classic database portal.

It is meant to replace UNData, which was fine for humans who like menus and filters, and less suited to a world where people increasingly ask chatbots for answers first. The platform also supports the Model Context Protocol, or MCP, which gives AI systems a standard way to connect directly to outside data sources. The UN is not just making the data searchable; it is also keeping track of where each statistic came from so a result can be traced back to the original source.

That traceability matters. UNICEF said a benchmark of six large language models across more than 133,000 answers to questions about global development indicators produced an average accuracy score of 21.2%. In that test, about three in five responses did not even include a usable number, and when the same questions were asked again about two days later, models that gave a number both times matched only about half the time. The study is still a working paper, not yet peer-reviewed, but the message is already blunt: AI is getting asked for authoritative data long before it is good at serving it.

UNICEF is seeing the shift in its own traffic. Its data site gets more than 6 million visits a month, and clicks from ChatGPT answers rose 67% year over year between January 1 and September 14. Those referrals made up 6.4% of all sessions this year, while UNICEF estimates AI assistants now drive about one in 10 visits overall. The pressure is obvious enough that the UN says 26 of its entities have signed on, with data from nearly 20 available at launch, and that it wants 80% of the system’s statistical datasets on the platform by 2027.

Google.org put in $2 million for capacity-building and technical support to get the core infrastructure in place. Google says the system is hosted on a UN-governed instance and is meant to be maintained and scaled by the UN itself over time. The company also showed how an AI agent could pull together UN indicators, build dashboards, charts, and written analysis, and even generate an infographic around the impact of PEPFAR in Africa. Useful, sure. But the warning label stays on: an AI can fetch the right numbers and still get the story wrong.