gldb is made by GHOST Projects, Inc., a one-person company run by Will Bryan. This page says what the service actually does with your data. Questions go to support@gldb.legal.
When you use the API, the Claude connector, or the Word add-in, gldb receives what you send it: case citations, case names, questions about them, and, when you run a citation check, the text of the document being checked. That content is used to answer your request.
Each request is also logged with a timestamp and a one-way hash of the access key that made it. The key itself is never written to logs.
That’s the whole list. The website sets no cookies and runs no analytics. We don’t collect names, emails, or payment details through the service itself; keys are issued by hand.
Passages of court opinions, and sometimes excerpts of your query, are sent to Anthropic’s Claude API to run the analysis. Anthropic doesn’t train on API data by default. The opinions themselves are public records, pulled from sources like CourtListener and the Caselaw Access Project.
Analysis results and usage logs live in a Postgres database in the United States. The service runs on Modal, also in the United States.
Document text you submit for a citation check is processed and returned. It isn’t kept.
The add-in stores your access key on your device, in Word’s web storage, so you don’t have to retype it. It’s sent only to the gldb API with your requests. Removing the add-in or clearing Office’s cache deletes it.
We don’t sell your data and we don’t share it for advertising. Your queries and documents aren’t used to build anything for anyone else.
Email support@gldb.legal to rotate or revoke a key, or to have the logs tied to your key deleted.
Changes get posted here with a new effective date.