Berkeley researchers use AI to make millions of local laws searchable

A University of California, Berkeley team has built an open-access database of millions of local ordinances across all 50 US states, using AI to turn a sprawling archive of legal documents into searchable data. The work could help researchers and the public compare rules that are usually scattered across thousands of local sources.

Berkeley BAIR Watch analysis

What happened

The database draws on nearly 10,000 documents, totalling about seven million pages, according to Bioengineer.org’s report. The team used LightOnOCR, a vision-language model, to extract text and structure from the documents, then used OpenAI models to label samples and organise the laws into categories. The resulting corpus is intended to support searches and comparisons across local regulations.

The report says the project was published online in June and that a paper describing the work has been accepted to NeurIPS, where the team is due to present it in December. It identifies Diag Davenport, an assistant professor at Berkeley, as the project lead.

Why it matters

Local rules shape everyday decisions about housing, building, business and public space, but they are difficult to compare when the source documents are fragmented and often hard to search. A national corpus could let researchers look for patterns across jurisdictions and help residents find relevant rules without trawling through piles of PDFs.

It may also make it easier to build tools for specific needs, such as comparing building requirements between neighbouring cities. The database is a research resource, not a guarantee that an AI chatbot built on it will give complete or legally reliable advice.

Our read

This is a useful application of AI to an unglamorous bottleneck: making public records readable at scale. The payoff is not simply a larger pile of searchable text, but the chance to ask better questions about how local rules differ and who they affect. For anyone using a tool built on the data, the underlying ordinance and its current status still matter. Law has never been famous for staying put.

What to watch

• Whether the researchers publish evaluation details on the corpus’s coverage and extraction accuracy.
• Whether journalists, researchers and local communities build useful tools from the open dataset.
• What findings emerge when local rules can be compared across jurisdictions.

Discussion spark: If local laws become searchable at national scale, what should come first: tools that help people find the rules that apply to them, or research comparing how those rules differ?

Independent WittyWires tracker for public updates about Berkeley BAIR. Not affiliated with or endorsed by Berkeley BAIR; this is not an official account.