Data License
How UnGovr data may be used, and which terms apply to which dataset.
UnGovr publishes two kinds of dataset under two different sets of terms. Most of what we publish is openly licensed. The AI and scraping law corpus is licensed separately, by agreement.
| Dataset | Terms |
|---|---|
| Government entities, geographic boundaries, transparency law records, California Grand Jury reports, and the rest of the Open Data API | CC BY 4.0, with attribution to UnGovr |
| The AI and scraping law corpus, including its bulk export | UnGovr Data License (non-exclusive, by agreement). Terms are set out below. |
UnGovr Data License (non-exclusive, by agreement)
This applies to the AI and scraping law corpus: the per jurisdiction records, the coverage index, the domain index, and the bulk JSONL export served from https://data.ungovr.org.
No license is conveyed by receipt of a file. Holding a copy of an UnGovr AI-law record, whether it reached you through the API, an export file, or a third party, does not grant you rights to use it. Rights are granted only under a written agreement with UnGovr.
What is licensed, and what is not
The underlying law is public. Statutes, regulations, bills, and court decisions belong to the public record, and nothing here claims otherwise. You are free to read, cite, and rely on primary sources directly, and every record we publish names its source so that you can.
What is licensed is our research: the selection of instruments, the verdicts and access matrices, the written summaries, the jurisdiction structure and slugs, the freshness and confidence metadata, and the compiled corpus as a whole.
Rights holder
UnGovr, Santa Barbara, California.
Attribution
Where use is licensed, attribute as UnGovr AI-law research, https://www.ungovr.org. Attribution does not substitute for a license, and it does not create one.
Non-exclusive
Licenses are granted on a non-exclusive basis. Licensing the corpus to one party does not restrict us from licensing it to another, and no licensee holds an exclusive position in any market or jurisdiction.
Research summary, not legal advice
Every record carries a not-legal-advice notice, and the notice travels on the record rather than on the response, because a bulk consumer loses the surrounding context on ingest. It reads:
Research summary of published law. Not legal advice and not authorization to access any system.
Both halves matter. A verdict in this corpus describes what our research found in published law as of a stated date. It is not advice about your situation, and it is not permission to access anyone's systems. Access to a specific service remains governed by that service's own terms and by the law that applies to you.
Freshness is part of the record
Law changes, and a legal answer with no date attached is worth very little. Every record carries the date the research was last reviewed, a confidence level, whether we consider the record stale, and whether the research is still awaiting human review. A licensee presenting our data to its own users should carry that metadata through rather than strip it.
Validating and ingesting the corpus
The published JSON Schema is served at https://data.ungovr.org/v1/schema/ungovr.ai-laws-2.schema.json and needs no API key, so a pipeline can validate on ingest before any credential is involved. The corpus itself is keyed. The schema is versioned: a consumer that meets a major version it does not recognise should refuse loudly rather than read the payload as best it can.
Licensing enquiries
To license the AI and scraping law corpus, or to ask which terms cover a dataset, write to api@ungovr.org.