DRAFT — not legally published. Counsel review pending. Do not rely on this text as a published policy.
Trust Center
Security, privacy and safety practices, plus the contacts that reach a responsible human.
PHASE B — DOCUMENT 8: SECURITY / TRUST CENTER
Status: draft for review. Canonical route: kidchat.io/trust.
Who operates KidChat
KidChat is a technology service operated by NOXON, LLC, a Texas limited liability company (Texas, United States). Español: KidChat es un servicio tecnológico operado por NOXON, LLC, sociedad de responsabilidad limitada constituida en Texas, Estados Unidos.
Contact: privacy@kidchat.io · safety@kidchat.io · security@kidchat.io · legal@kidchat.io
No further corporate information is published unless legally required. Notice address per jurisdiction: [COUNSEL REVIEW REQUIRED — PRIVACY CONTROLLER / RESPONSIBLE ENTITY ADDRESS]
Primary message: "Built for children means held to a higher standard."
A. TRUST CENTER STRUCTURE
| Route | Contents |
|---|---|
| /trust | Entry page, the message above, links to everything below |
| /trust/privacy | Parent Privacy Policy / Aviso de Privacidad (doc 2) |
| /trust/children | Children's Privacy Notice, both layers (doc 1) |
| /trust/safety | Child Safety Policy (doc 5) |
| /trust/ai | Responsible AI / AI Limitations (doc 6) |
| /trust/security | This document |
| /trust/parents | Parent Rights & Controls (doc 7) |
Legal copies may also live at kidchat.io/legal/privacy, /legal/terms, /legal/cookies. /trust is the parent-facing entry point.
B. WHAT WE CAN SAY TODAY
- Data is encrypted in transit (HTTPS everywhere) and at rest by our managed database provider. (C-17) — wording must match the provider's own documentation.
- Access to a child's conversations requires the account holder's authentication plus Parent PIN or re-authentication, and each access is recorded. (C-12)
- Database access is restricted per account by row-level security: one family's data is not reachable from another family's session.
- Secrets and provider keys are held server-side only and are never exposed to the browser.
- Gateway prompt/content logging is off for child traffic; the AI cost ledger stores no conversation content. (C-18)
- No advertising SDKs, no advertising identifiers, and no third-party behavioral analytics run in Kid Mode. (C-09, C-10)
C. WHAT WE WILL NOT CLAIM
No certification or audited-compliance claim of any kind: no SOC 2, no ISO, no PCI, no "COPPA certified", no "GDPR compliant", no penetration-test badge, no 24/7 monitoring, no "your child's data is never shared". If we obtain one, we will publish the report date and scope, not a logo.
D. INFRASTRUCTURE AND SUBPROCESSORS
A current subprocessor list is published and dated at /trust/security: hosting/CDN, managed database and authentication, AI gateway and approved model providers, transactional email. Each must clear vendor review before touching child data: retention, training, logging, subprocessors, security, DPA, data location and transfer mechanism, deletion, incident notification. [TECHNICAL CONTROL REQUIRED — vendor files and dated subprocessor page]
Voice is off by server-side feature flag: microphone and audio endpoints are unreachable, and no dormant microphone functionality that records or transmits audio is exposed. (C-16) Re-enabling requires provider review, contract, retention review, consent analysis, safety review, and a privacy-copy update.
E. BACKUPS AND DELETION
Backups exist for disaster recovery. Deletion from active systems happens as soon as reasonably practicable; the target backup purge window is 30 additional days, pending measurement. [TECHNICAL VALIDATION REQUIRED — BACKUP RETENTION] Restored data is used only for disaster-recovery or security purposes, and deletion requests are re-applied where technically required.
F. INCIDENT RESPONSE
Report a vulnerability or suspected incident to security@kidchat.io. We acknowledge reports, investigate, and notify affected parents and regulators where the law requires it — in Mexico under the LFPDPPP breach-notification duty, in the United States under applicable state breach laws. We do not promise a notification time we have not committed to operationally. [TECHNICAL CONTROL REQUIRED — incident runbook + contact rotation]
G. SEGURIDAD — español (México)
- Ciframos la información en tránsito (HTTPS) y en reposo mediante nuestro proveedor de base de datos administrada.
- El acceso al historial requiere autenticación del titular de la cuenta y NIP, y cada acceso queda registrado.
- Aislamiento por cuenta mediante seguridad a nivel de fila.
- Sin SDK de publicidad, sin identificadores publicitarios y sin analítica conductual de terceros en el Modo Niño.
- No afirmamos ninguna certificación (SOC 2, ISO, PCI) ni cumplimiento auditado.
- Incidentes: security@kidchat.io. Notificamos a las personas afectadas y a la autoridad cuando la ley lo exige.
Voice (Alpha) — DRAFT [COUNSEL REVIEW REQUIRED — SPEECH PROVIDER / AUDIO CONSENT]
Voice is optional and off by default. It works only for Alpha families, and only after a parent turns on "Allow voice" for a child in Parent Mode. When it is on, the child's spoken words are sent through OpenRouter to AI speech models to turn them into text, and the buddy's answers are sent the same way to turn them into speech. KidChat does not store audio: each clip is used once and discarded. The text goes through the same Privacy Shield and Safety Gate as typed messages. A parent can turn voice off at any time.
Who operates KidChat
KidChat is a technology service operated by NOXON, LLC, a Texas limited liability company.
KidChat es un servicio tecnológico operado por NOXON, LLC, sociedad de responsabilidad limitada constituida en Texas, Estados Unidos.
NOXON, LLC — Texas, United States. hello@kidchat.io
