Small data footprint, plain-language choices
Privacy Policy
This policy explains what Talking Robot processes, what it deliberately does not collect, and how voice requests are handled.
Last updated: 15 August 2026
Kenzi does not store microphone recordings in Talking Robot’s database or file storage. Audio must still be transmitted to our AI service provider to understand the child and create a response.
1. Privacy by design
Talking Robot is operated by Kenzi.ai and is designed for children and families. We aim to collect the minimum information needed to run the game, protect the service, calculate battery use, and improve reliability.
The app does not ask a child to create an account or provide a name, email address, phone number, home address, school, contacts, precise location, or profile photo. It does not include public profiles, advertising, or a social feed.
2. Information the app processes
The service processes limited device and operational information, including:
- A random device/installation identifier and protected authentication token. The installation identifier is stored as a one-way hash on our server.
- Platform, app version, language/locale, timestamps, and last activity time.
- Battery balance, battery transactions, purchase validation records when purchases are introduced, and related anti-fraud information.
- Technical interaction metrics such as audio duration, token counts, reply length, success/failure status, and diagnostic correlation identifiers.
- Standard security logs, which may include IP address, user agent, request time, route, and response status, for security, abuse prevention, and reliability.
3. Voice recordings and AI processing
The microphone activates only when the child or parent starts a talk interaction and the device grants microphone permission. The recorded clip is transmitted securely to the Talking Robot API and then to OpenAI’s API services to convert speech to text, generate a child-aware reply, and create the robot’s spoken audio.
Kenzi does not save the microphone recording to its database, local server files, or cloud object storage. We process it in memory for the request and discard it from our systems after the request completes. We also do not store the full transcript or robot reply in our persistent interaction database.
OpenAI processes submitted data as our AI service provider under its applicable business/API data terms and security controls. Provider-side handling may be subject to those terms and lawful security or abuse-monitoring requirements. Families should not say names, addresses, contact details, or other sensitive personal information to the robot.
4. Short-term conversation memory
To make a short exchange feel like a conversation, up to two recent child-and-robot turns may be held in volatile server memory for up to 15 minutes. This memory is linked to an installation and conversation identifier, is not written to the conversation database, and disappears when it expires or the service restarts.
5. Why we use information
We use the limited information described above to:
- Register and recognize an app installation without creating a child account.
- Understand a voice request and return an AI-generated spoken response.
- Apply child-safety checks to incoming and outgoing content.
- Maintain battery balances, validate purchases, prevent duplicate charges, and investigate fraud.
- Protect the API, diagnose errors, monitor availability, and improve product performance.
- Meet legal obligations and respond to valid requests from authorities where required.
6. When information is shared
We do not sell children’s personal information and do not use it for behavioral advertising. We share limited information only with service providers needed to operate the product—such as OpenAI for speech and response processing, hosting/security providers, Bunny CDN for static website files, and Apple or Google for store purchases—or when required by law.
Service providers are permitted to process information only for their contracted role and under their applicable data-protection commitments. We may also disclose information to protect a child, user, Kenzi, or the public from a credible safety or security threat, where legally permitted.
7. Retention and security
Voice recordings are not retained by Kenzi. Temporary conversation memory expires as described above. Device, battery, purchase, security, and technical records are kept only as long as reasonably needed to operate the service, meet financial or legal obligations, resolve disputes, and prevent abuse, after which they are deleted or anonymized where practical.
We use transport encryption, protected credentials, access controls, rate limiting, hashed identifiers, and other safeguards appropriate to the information. No online service can promise absolute security, so we continually review and improve these controls.
8. Parents’ choices and rights
A parent or guardian can deny or revoke microphone permission in the device settings; offline building and reactions can still be used. Removing the app may remove local preferences, but some server-side battery and transaction records may remain where needed for purchase restoration, fraud prevention, or legal compliance.
Depending on where the family lives, a parent may have rights to request access, correction, deletion, restriction, or a copy of personal information. Because there is no account, we may request limited information needed to verify the relevant installation and the requesting adult’s authority.
9. International processing and policy changes
Our providers may process data in countries other than the family’s country. We use appropriate contractual and technical measures where required. We may update this policy as the product, providers, or law changes; the current date will always appear at the top of this page.
Privacy question or parent request?
Reach the Kenzi team through our official website and mention Talking Robot.
This information is written in plain language for families and is not a substitute for professional legal advice.
