Trust center
This launch summary explains the operating promises built into Calm Answers. Final public policies will receive legal and child-safety review before broad release.
Calm Answers checks an exact canonical cache before generating a new response. New answers use structured output and current web sources when facts may have changed or the topic is high-stakes. Answers can be challenged, reviewed, replaced with a new version, or retired.
Private and identifying questions are excluded from the public canonical cache. Sponsors receive only aggregate topic reporting—never raw questions, child activity, safety events, precise location, or personal history. Account history can be disabled and cleared.
The ban applies to every plan. Non-graphic health, puberty, consent, relationship, and abuse-help questions remain available. Restricted safety-event metadata is separated from general analytics and requires purpose-based access. No automated classifier makes a police report.
Parents create nickname-based child profiles and start an authorized-device Learning Mode. Child questions receive guided-learning behavior, no sponsorship, and a parent reveal-request path. Confirmed child explicit-content blocks create categorized alerts without raw wording by default.
At most one approved placement appears after the organic answer and sources. Premium, child, Learning, health, legal, financial, crisis, safety, and disputed contexts suppress sponsorship automatically.
Anyone can report an answer. Only authorized reviewers can publish a correction. Publishing creates a new answer version, retires the disputed version, and moves the canonical topic to the corrected response.