Privacy
Do not enter client names, dates of birth, addresses, case numbers, or other personally identifying information into search queries, follow-up questions, or export notes. Write at the level of a population or a practice problem instead of a person.
What leaves your browser
- Search terms go to OpenAlex to retrieve scholarly metadata, and to Crossref when a result has a DOI that needs verifying.
- SocialGPT questions, along with the titles and years of the sources returned for that search, go to OpenAI to generate the reply.
- Read aloud sends the text of a SocialGPT reply to OpenAI’s speech endpoint to produce audio. The audio is not stored.
What stays on your device
The working set behind assignment exports is kept in your browser’s local storage. It is not sent to a server, and clearing site data will remove it. There are no accounts, so nothing is tied to an identity.
Coming later
Saved research features will store only the account holder’s own library data, under authentication and row-level security. This page will be updated before any of that ships.
More on how results are produced is on the methodology page, and the FAQ covers what SocialGPT will and will not do.