Privacy Policy

Parsec Guardian · Last updated: replace with your date

Parsec Guardian is a parental-safety tool. A parent installs it on their own minor child's browser to be alerted when the child's AI chatbot conversations contain concerning content. This policy explains what the extension collects, why, and how it is handled.

Who this is for

Parsec is intended solely for a parent or legal guardian monitoring a device belonging to their own minor child. It should not be installed on any device without the right and authority to monitor it.

What the extension accesses

• Conversation text on supported AI chatbot sites (ChatGPT, Claude, Gemini, Character.AI, Poe, Microsoft Copilot, Perplexity, Grok, Meta AI, JanitorAI, Chai, Nomi.ai, DreamGen, Abby AI, Pi, Mistral, HuggingChat, DeepSeek, Doubao, and a self-hosted SillyTavern instance). The extension reads the on-page chat messages so they can be analyzed for safety.
• The site name and page URL of the chat, to label where activity happened.
• A pairing token stored locally on the device, which links it to the parent's account. It contains no personal data and can be revoked by the parent at any time.

The extension does not read pages outside the supported AI chat sites. Its only visible surface on the child's device is a notification when a moment is flagged — see below.

When a moment is flagged

Before a flagged moment is shared with the parent, the extension shows a notification on the child's own device letting them know it's being shared and offering an optional text box to add context (for example, that it was for schoolwork or research). The child's note, if any, is sent to the parent alongside the flagged moment. Adding context is optional: with or without a note — and whether or not the child has notifications enabled — the flagged moment is always shared with the parent after a short hold.

How the data is used

• Captured conversation text is sent over an encrypted connection to the Parsec backend, where it is analyzed by an AI classifier (Anthropic's Claude API) for a small set of safety categories: self-harm, predatory behavior, sexual content, violence, and parent-defined custom topics.
• Only when content is flagged is a short relevant snippet stored and shown to the parent. Text that is not flagged is analyzed and then discarded — full conversations are never stored.
• Approximate per-platform activity (which platforms were used, and when) is recorded so the parent can see where their child uses AI.

What is stored, and who can see it

Stored data — parent account details, flagged snippets, platform activity, and custom trigger settings — is kept in the Parsec backend database and is visible only to the parent who owns the account, through their password-protected portal. Snippets are short excerpts, not full transcripts.

Sharing

We do not sell personal data or use it for advertising. Conversation text is shared only with the AI classification provider (Anthropic) for the sole purpose of safety analysis. It is not used to train models.

Data retention and deletion

A parent can revoke a device (stopping all monitoring) and delete flagged snippets and custom triggers from the portal at any time. To delete an account and all associated data, contact us at the address below.

Security

Data is transmitted over HTTPS. API keys and database credentials are held only on the backend and are never included in the browser extension.

Children's privacy

Parsec is a tool operated by a parent or guardian on behalf of their own minor child, under the parent's authority and consent. Parents are responsible for using it in accordance with applicable laws in their jurisdiction.

Contact

Questions or deletion requests: replace@yourdomain.com.