United Nations Partners With Google to Build AI-Ready Global Data Platform

The United Nations has partnered with Google to deploy a centralized data infrastructure designed to make international statistical datasets accessible to artificial intelligence systems. The initiative, named the UN System Data Commons, updates the organization's legacy UNData repository with a platform built on Google's open-source Data Commons architecture. The platform introduces natural-language search capabilities and implements the Model Context Protocol (MCP), a standardized framework enabling automated AI agents to interface directly with external databases.

The deployment comes amid low reliability scores for major commercial artificial intelligence models querying global development statistics. In a benchmark study conducted by UNICEF that evaluated over 133,000 responses across prominent large language models, systems achieved an average accuracy rate of 21.2%, as first reported by TechCrunch AI. The research assessed OpenAI's GPT-4o and GPT-4o-mini, Anthropic's Claude Sonnet 4.5 and Haiku 4.5, and Google's Gemini 2.5 Flash and Gemini 2.0 Flash, according to João Pedro Azevedo, UNICEF's chief statistician.

UNICEF's analysis revealed that approximately three out of five model responses omitted explicit numerical values, frequently due to model hedging. Furthermore, when identical prompts were resubmitted two days later to the same software versions, the models provided consistent numerical outputs in only about half of the instances where figures were generated. The findings originate from an un-peer-reviewed UNICEF working paper targeted for academic journal submission, with the agency planning to publicly release its underlying code, data, and methodologies.

The effort to establish structured data feeds also addresses accelerating web traffic from generative AI applications to official international repositories. UNICEF's statistics portal, which receives more than six million visits monthly, recorded a 67% year-over-year increase in referral traffic from ChatGPT links between Jan. 1 and Sept. 14, Azevedo disclosed to TechCrunch AI. ChatGPT referrals currently account for 6.4% of overall portal sessions, while total traffic generated by AI assistants represents approximately 10% of all site visits.

To date, 26 UN entities have agreed to integrate their statistics into the Data Commons framework, with nearly 20 participating at launch. The United Nations aims to bring 80% of its global statistical datasets into the unified system by 2027. "We are orders of magnitude more advanced in scale, scope, and flexibility, connecting for the first time across so many agencies across the UN system," said Shantanu Mukherjee, acting director of the UN Statistics Division. "And [we are] taking this moment to also make our data AI-ready."

Google.org contributed $2 million in technical guidance and capacity-building funding to help establish the core infrastructure. Prem Ramaswamy, lead for Google's Data Commons initiative, noted that the platform operates on an instance governed directly by the UN. Google has utilized a train-the-trainer strategy to allow internal UN teams to eventually maintain, run, and scale the system independently, Ramaswamy told TechCrunch AI.

Originally established by Google in 2018 to index public statistics, Data Commons integrated support for MCP last year to allow AI agents to query external figures alongside source metadata. Under the UN implementation, retrieved data points maintain clear lineage to their original sources to preserve verification standards. While automated tools can use MCP to synthesize complex datasets into charts, dashboards, and reports—such as analyzing the impact of the U.S. President's Emergency Plan for AIDS Relief across HIV infection, mortality, and life expectancy metrics—Ramaswamy emphasized that human oversight remains critical. "Because models can misinterpret nuance, a human should always review the outputs before citing or publishing them," he noted.