Today, we are proudly announcing a new collaboration with the Government of Brazil to make public data more accessible, trustworthy, and usable in the age of AI.
Starting this month, the Open Knowledge Foundation (OKFN)’s Tech Team is working with the Office of the Comptroller General (CGU) to prototype an integration with open data portals that enables users to ask natural-language questions and receive responses traceable directly to the underlying datasets.
For decades, the Brazilian government has been a leader in open data, proactively making large datasets available to its citizens. However, until now, extracting information from these datasets has required citizens to have a certain level of expertise, as well as taking up additional time and effort. This pilot is exploring ways to make public data more accessible and usable. Citizens will be able to ‘talk’ to public data and receive AI-generated answers traceable to their data source, and accurate, transparent, and verifiable. This will contribute to reinforce trust in public institutions, help fight misinformation, encourage data reuse and ensure that research, policy and news are based on verified, up-to-date data.
Traceable AI answers
The technical solution we are prototyping uses a Model Context Protocol (MCP), a technical bridge that enables AI tools to connect directly to trusted data sources. In this pilot, the MCP will be implemented for CKAN, a digital public good and the world’s leading open source data management system that powers most of the world’s public data portals. (CKAN was created at OKFN and is turning 20 in 2026!)
This means that instead of generating answers from a vast amount of generic data, the system will be able to:
- Pull information directly from official datasets
- Provide responses that are explicitly grounded in those sources
- Link users back to the original dataset for verification and additional information from its metadata.
- Reinforce digital sovereignty and autonomy.
In short, AI answers become traceable and therefore more reliable.
Working with a pioneer in public data and digital public goods for an exponential effect
The collaboration with the Government of Brazil, the largest and most sophisticated open data ecosystem in the region, will open possibilities for other governments, international institutions, and the private and social sectors to follow the lead.
The pilot is taking place under the umbrella of the Digital Public Goods Alliance (DPGA) to become a reproducible and scalable reference of responsible and transparent AI integration with open data infrastructure, with the potential to be shared broadly across institutions in Brazil and across the growing number of countries officially engaging with the alliance, and beyond.
This solution will also have the potential to be highly customisable, with technology-agnostic components that fit together seamlessly – meaning it will not be a locked-in solution limited to CKAN-powered portals, or to just a few languages, or even to a specific LLM. Our aim is to genuinely address the issue of traceability for AI-generated responses and to contribute to trust in public data.
“This is a small technical step built on top of years of hard work by access to information activists, dedicated public servants and brilliant technical communities that made open data a priority for governments and understood that public data is critical infrastructure for a nation. Open access to public information is a precondition of strong democracies. It is good governance. And Openness is not only making the information available only for some with technical skills. Meaningful open data is accessible open data for any citizen. That is what this alliance is going to deliver,” said Renata Ávila, CEO of the Open Knowledge Foundation.
“Brazil has invested for years in building a robust open data infrastructure,” said Otavio Neves, Open Data Policy Coordinator at CGU. “Through this initiative, we aim to expand the reach of Brazilian public data further and contribute to an innovative approach to strengthening democracy.”
A community-oriented effort
We aren’t building this in a vacuum. True transparency in AI must be a collective effort, which is why we are opening our development process and technical consultations to the global open data and open knowledge communities, approaching this issue with an eye to the ethical risks involved while also presenting an opportunity for collective learning.
Last March, at The Future of Open Data (an official Open Data Day 2026 event), we initiated a public discussion with key figures from the community, including all CKAN co-stewards and contributors, to explore the future of open data and share our ongoing projects. You can watch this conversation here.
We also organised a technical consultation for engineers and developers, where we discussed different architectures that the community has been exploring. We are aware of initiatives in France, the United States and Italy that are moving in the same direction with MCPs, and we intend to set up ground for collaboration. This meeting was documented here and prompted an open and honest technical reflection from our team.
Join the conversation and contribute with your expertise!
We will share and discuss the entire development process with the community at the Open Knowledge Forum. You are welcome to join the discussion and share your thoughts, as well as more initiatives bridging AI and open data. You can also email us at info@okfn.org anytime.
Acknowledgement
We are grateful for the Patrick J. McGovern Foundation’s (PJMF) generous support and our continued partnership in enhancing digital literacy and investing in AI for the public good. Learn more about its charitable programmes here.
