Privacy Policy

Kaze is an AI chat companion for children, operated by Kaze App. This policy explains what personal data we collect, how we use it, who we share it with, and the rights you have over your family's information.

What information we collect

We collect different types of information depending on who is using the service. From parents and guardians, we collect: name, email address, account settings, and IP address. From children (through the parent account), we collect: name, age, learning profile (interests, learning style, reading level, languages spoken), and conversation history with the AI companion. Automatically, we collect: browser type, device information, and session data used for kid-device linking. We do NOT collect: location data, advertising identifiers, or social media profiles.

How we use information

We use the information we collect to: provide the AI chat experience for children, enable parental oversight and review of conversations, generate activity summaries for parents, perform safety monitoring and content moderation, and manage accounts and authentication (including magic link emails). We do NOT sell personal data. We do NOT use personal data for advertising. We do NOT build behavioral profiles for marketing purposes.

AI service providers

Replies are written by Anthropic's Claude models. When your child sends a message, we transmit to the provider that writes the replies: the child's message with personal details removed, relevant conversation context, and chat instructions built from the child's age, interests and learning style, so responses are age-appropriate. The same provider writes the chat summaries you see. Safety checks are made by TypeSafe's Jev model, which reads the message as your child wrote it, because it has to judge exactly what was written; TypeSafe returns scores and never writes a reply. When TypeSafe can't decide, Anthropic's Claude classifier, on Kaze's own account, reads the message as your child wrote it too. The same classifier checks, as written, the other texts Kaze sends to an AI: your child's interests, a grown-up's notes and messages, the chats a summary is written from, and class material. None of these providers uses this data to train its models. Anthropic deletes the data it receives within 30 days, except content its automated safety systems flag, which it can keep for up to 2 years (and the safety scores for up to 7). Anthropic works for Kaze under a data processing agreement, and may process the data outside the European Union, including in the United States, under the EU Standard Contractual Clauses. We also use an error-monitoring service (GlitchTip) that may receive technical diagnostic data but never conversation content.

Study projects (when your family has them): the class name and description you or your child type, and the text of the class PDFs you attach, go to Anthropic's Claude Opus model, on Kaze's own account, to propose the class's topics and objectives. A scanned PDF with no text goes whole, so the model reads its page images and writes down the course content. This happens whichever provider writes your child's chat replies. The text passes the same safety checks as a message first, with personal details removed; a scan's transcription is checked before anything uses it. The PDF files themselves are stored encrypted on Kaze's servers, and deleted when you delete them, the class, the child or your account. When a class is finished, its PDFs and the text read from them are deleted 90 days later; the class page says on which day, and the topics, progress, report and exams stay.

Practice exams in study projects: the questions are written by Anthropic's Claude Opus model from the class and its PDFs, and checked by Anthropic's Claude Sonnet model, on Kaze's own account. Your child's answers pass the same safety checks as a message, and go to Anthropic, with personal details removed, to be graded. An answer the safety checks flag is held for you, as a message is, and isn't graded until you decide. The answers are stored encrypted, and deleted with the class, the child or your account.

Study sessions (when your family has them): in a study session the tutor, Anthropic's Claude Sonnet model on your family's key, reads the class material (the objectives and the text of the class PDFs, with personal details removed) and the session's messages, with personal details removed as in a chat. Each message your child sends in a session also goes to Anthropic's Claude Sonnet model on Kaze's own account, with personal details removed and with the tutor's last reply, to check the answer before the tutor replies. Only the result of that check is kept (whether the answer was right, and the likely reason if not), never the text.

Data storage and security

Your data is stored on secured servers. Conversations are stored in our database so that parents can review them at any time. All data in transit is encrypted (the same security used by banks and online stores). You can sign in with a magic link, with Google, or with a password; a password or parent PIN is stored only as a one-way hash (bcrypt), never in plain text, so nobody at Kaze can read it. Child sessions are created through device-linking codes and are time-limited.

Data retention

For active accounts, we retain data for as long as the account remains active. When a child profile is deactivated, the associated data is retained for 90 days and then permanently deleted. If you request full account deletion (available in Settings > Data & Privacy), all your data is permanently removed. You can also download all your data at any time in JSON format from the same settings page.

Parental rights

As a parent or guardian, you have the right to: review your child's personal information and full conversation history, download all of your family's data (Settings > Data & Privacy > Download my data), delete any child's profile and all associated data (Settings > Children > Delete), delete your entire account and all associated data, and withdraw consent at any time by deleting your account. For residents of the European Union, you additionally have the right to access, rectification, erasure, data portability, and objection under the GDPR (the European Union's data protection law). To exercise any of these rights, use the in-app settings or contact us at privacy@kazeapp.com.

Children's privacy

Kaze is designed for children under parental supervision, in accordance with COPPA (Children's Online Privacy Protection Act). We require verifiable parental consent before any child can use the service. We collect only the information necessary to provide the service. Children cannot share personal information publicly through Kaze — all conversations are private between the child and the AI, visible only to the child and their parent or guardian. AI safety filters monitor all conversations for inappropriate content.

Cookies and tracking

We use essential cookies only, for session management and authentication. We do not use advertising cookies, analytics tracking cookies, or any form of third-party tracking. There are no social media trackers, no remarketing pixels, and no cross-site tracking of any kind.

Changes to this policy

We may update this privacy policy from time to time. If we make material changes, we will notify you by email. The version number and effective date are shown at the top of this policy so you can always tell when it was last revised.