AI Processing and Output Disclosure

Your granular choices

• Basic processing of direct feature input.
• Tasks and plan context.
• Check-in context.
• Health and wellness context.
• Calendar availability, separately from titles and descriptions.
• Profile preferences and uploaded content.

All choices start off. Sensitive choices do not imply one another. Detailed calendar sharing requires availability sharing; without detail authorization, occupied time is represented as “Busy time.” Consent is versioned, expires, and can be revoked in Settings. Revocation invalidates permits, queued context, and caches before another request can leave Daybreak.

Provider handling

The gateway sends only categories required for the selected feature through a short-lived, user- and purpose-bound authorization. The OpenAI API does not use API business data for training by default; provider abuse-monitoring retention may apply unless Daybreak's approved project has stricter controls. Production use remains disabled until contract, data-processing, retention, region, security, and owner approvals are recorded.

Output safety

Outputs are probabilistic and can be inaccurate, incomplete, biased, or inappropriate. Daybreak validates structure and allowed actions, but you must review output. AI cannot change consent, legal acceptance, measured health records, subscription state, or another user's data, and must not diagnose or prescribe medication. For urgent danger, contact local emergency services; Daybreak cannot contact them for you.