Analysis happens only after your review, disclosure, and request.
Aureni uses a third-party artificial-intelligence service, OpenAI, Inc. (San Francisco, USA), to generate reflections. Before the first analysis, the app shows a disclosure that lists exactly what will be sent and to whom, and nothing is transmitted until you choose “Allow.” You can reopen that disclosure from the capture screen at any time, and you can withdraw AI consent in the app’s Privacy screen.
When you request an analysis, the following is sent from your device to the Aureni analysis API (operated by Arcturne LLC) and, after an automated safety check, to OpenAI’s API:
- the text of the moment you reviewed and edited on the capture screen;
- the alias you chose for the person;
- your stated goal and the standards you picked;
- the support style you selected for that reflection; and
- a pseudonymous installation or account identifier used for abuse prevention, which is not your name, email, Apple ID, or Google Account ID.
Aureni does not attach raw screenshots or audio, account credentials, structured name or email fields, contacts, precise location, or your encrypted history to an OpenAI analysis request. The single reviewed moment and the context listed above are included. Text you submit, including an alias or a quoted conversation, can contain names, contact details, health information, or other personal information. Automated redaction is incomplete; remove anything you do not want to share before submitting.
Aureni requests that the response not be stored as a reusable OpenAI response by setting storage off for the request. Under OpenAI’s API data usage policies, content sent through the API is not used to train OpenAI’s models, and OpenAI may retain API data for a limited period for service operation, safety, and abuse monitoring under its applicable API data controls unless Arcturne’s production account is approved for more restrictive retention.
Aureni does not write the raw analysis input to its application database. It stores a one-way input hash for safe retry handling and may temporarily retain the generated response for up to 24 hours so an interrupted request can be returned consistently. The temporary response payload is then removed.
AI output can be inaccurate, incomplete, or inappropriate for a particular situation. Aureni is not designed to infer another person’s private thoughts or intentions and does not treat possibilities as facts.