Data.gov review: the best US government datasets for AI projects
Answer in brief: Data.gov is the US government's open data catalog, listing 532,480 datasets as of 2026-11-09. It's the authoritative place to find public-sector, geospatial, economic and civic data to fine-tune, ground or evaluate AI systems for government, policy and domain-specific work. It's free and organized by agency, with usage metrics and a next-generation catalog at catalog.data.gov. Formats and quality vary widely, so expect to do your own ML preparation.
Best for: official public-sector, geospatial and economic data for domain AI.
Independent research from the official Data.gov homepage, checked 2026-11-09. Not hands-on testing.
What it does for AI datasets
Data.gov is run by the GSA's Technology Transformation Services. It gathers federal open data with tools to browse the most viewed and most recently added datasets, datasets by organization, and a geospatial filter. A metrics section breaks down page views, downloads and external link clicks by agency. Under the SHARE IT Act, Data.gov also lists federal source code repositories.
What to grab here, by task
- Geospatial and Earth AI: the geospatial collection.
- Tabular forecasting and policy models: economic, health and civic datasets from federal agencies.
- Retrieval grounding and evaluation: official government records to ground or test assistants built for the public sector.
- Code: the federal source code repository listings.
Who it's built for
Civic tech teams, researchers, public-sector AI builders and anyone who needs data from an official source.
Where it shines
- Huge official coverage: 532,480 datasets.
- Organized by agency, so you can see where each dataset comes from.
- Transparent metrics on what people actually use.
Honest drawbacks
- Formats and quality vary widely.
- Little is packaged for ML: no ready-made splits or loaders.
- The catalog is in transition, and Data.gov asks users to try the next-generation version.
Pricing and licenses
Free. Most federal data is open, but check each dataset's terms.
Best alternatives
- Google Dataset Search: search government and research repositories together.
- Registry of Open Data on AWS: NOAA, NASA and NIH data at cloud scale.
- Zenodo: research datasets with DOIs.
FAQ
How many datasets are on Data.gov?
532,480 datasets were listed on 2026-11-09.
Is Data.gov data free to use?
Access is free. Most data is open, but check each dataset's terms.
Who runs Data.gov?
The GSA's Technology Transformation Services.
Sources
- Data.gov
- catalog.data.gov
- Data.gov metrics
← Back to the ranked list of the best AI datasets
Named on this page as a provider? Join the network for a linked profile, Visit buttons and featured placement.