Project status
KidChat is in closed alpha. This page shows what already works, what is being built and what has to be finished before real children use the product.
Working today
- Real AI conversations for Ask, Learn, Discover and Create
- Real parent accounts: sign up, sign in, password recovery
- Children stored per family, with data isolated between families
- Screen time: daily limits, allowed windows, breaks and PIN extra time
- Session recaps with topics, highlights and safety indicators
- Emails from thinkfirst@kidchat.io on a verified domain
- Age range limited to 6–12, enforced in the app and in the database
- Always-visible child notice: a grown-up can read these conversations
In build
- Safety Gate: one single server path for every child AI request
- Privacy Shield: sensitive details removed before anything reaches the AI
- Output safety filter and a verified helpline directory
- Approved provider and model allowlist with fail-closed behavior
- AI cost ledger that never stores what was said
- 30-day retention, parent-initiated deletion and data export
- Parent access audit log and safety escalation queue
Deliberately off
- Voice: dictation, read-aloud and voice previews are turned off on the server until a speech vendor contract exists
Blockers before real children
- Verifiable parental consent flow (assessed separately for Mexico and the United States)
- privacy@kidchat.io monitored by a responsible human before any real child data
- Provider no-training and retention commitments evidenced in writing
- Counsel: notice address for the responsible entity per jurisdiction
- Counsel: governing law and dispute mechanics per jurisdiction
The full documents live in the legal center and the Trust Center. Every screen is listed in the site map.