The free plan is supported by a narrow, explicit permission to learn from de-identified text patterns. Your private archive remains yours.
WHAT MAY BE USED
De-identified patterns made from eligible text
How people describe time, relationships and events
Ways conflicting memories are clarified
Patterns that help improve memory organization and model responses
Never a copy of your private memory record
NEVER INCLUDED
Identity and sensitive parts of a life
Your photos, video, raw audio or raw transcripts
Names, contact details, precise addresses or direct identity
Children's data or biometric voice and face material
Trauma, abuse, health, sex life, religion, politics, finances or legal matters
The rules
You choose it clearly.The free-plan permission has its own checkbox and versioned consent record. It is not hidden inside general terms.
You can revoke future use.Your written archive stays available. Subsidized voice and new AI-derived work pause unless you restore consent.
Prepared data is withdrawn.Eligible derivatives that have not been released are marked for withdrawal. Data already used to train a released model cannot be removed from that model.
Future plans need a fresh choice.There is no paid plan today. If private, non-training-supported accounts open later, switching will be explicit.
Every use keeps provenance.Consent version, source class, de-identification version and withdrawal state are recorded so use can be audited.