Noterva Assistant — Privacy Policy

Privacy policy for Noterva Assistant — Summarize, Explain, Translate on Chrome, Edge and other Chromium browsers. Effective 24 July 2026.

In one line

Noterva Assistant does its work on your own computer. By default the AI runs locally and nothing you select ever leaves your machine. The only personal data we hold is what you give us if you choose to create a Noterva account — your email and password — used to sign you in and unlock Premium.

Our principles

• On-device by default. The standard mode processes your selection locally; it is never transmitted.
• Data minimisation. We process only what a stated feature needs.
• Purpose limitation. Data used for one purpose is not quietly reused for another.
• No sale, ever. We do not sell or rent your data, to anyone, for anything.
• Your rights. You can ask to view or delete any data held about you at any time.

What we do not collect

• No browsing history. We do not record or transmit the pages you visit.
• No page content beyond your selection. Only the text you explicitly select, only when you invoke an action.
• No analytics or telemetry. No usage tracking, no behavioural profiling, no advertising SDKs.
• No remote code. All logic ships inside the extension; nothing is fetched and executed at runtime. The downloaded AI models are data files, not code.

What this extension does

You select a passage and ask Assistant to summarize, explain, translate, rewrite, shorten, lengthen or fix its grammar. By default this happens entirely on your own computer — small AI models are downloaded once and run locally in your browser, and your selected text is never transmitted. You can optionally choose a cloud engine instead: Groq is free, and OpenAI, Claude and Google Gemini are Premium. In every cloud case you supply your own API key, and your selected text goes directly from your browser to the provider you chose — it never passes through our servers. Co-Pilot mode simply builds a prompt from your selection, copies it to your clipboard and opens your own AI so you can paste it there. Creating a Noterva account is optional and needed only for Premium.

Data that leaves your device

What	When	To whom	Why
Nothing — your selected text	Default on-device mode	No one	The AI runs locally; your selection is never transmitted.
AI model files (download only)	Once, on first use of on-device mode	huggingface.co (or the hf-mirror.com fallback)	To download the local AI models. No user data or selected text is sent in the request.
Your selected text + your own API key	Only if you choose a cloud engine, when you act on a passage	The provider you configured (Groq / OpenAI / Anthropic / Google)	To generate the result you asked for, under that provider’s policy. Never routed through us.
Your account email and password	When you create or sign in to a Noterva account	Noterva account server (noterva.com)	To authenticate you and unlock Premium. The password is stored only as a secure one-way hash, never kept in the extension.
A random device ID (and any offline activation code)	On activation and periodic licence re-checks	Noterva licensing (noterva.com); ajayceindia.com to check an offline code is unused	To validate your Premium licence and enforce the per-account device limit. The device ID is random and not derived from your hardware.

Your API keys are stored only in your browser and are never transmitted to Noterva. In free co-pilot mode, no selected text is sent to us or to any provider by the extension — you paste it yourself into the AI site you opened.

Browser permissions

• contextMenus. Add the right-click “summarize / explain / translate” actions.
• storage, unlimitedStorage. Save your settings, licence/account state and (optionally) your own API key locally; the on-device AI models are about 1 GB and need the larger quota.
• activeTab, scripting. Read the passage you select and show the result on the current page, only when you invoke an action.
• offscreen. Run the on-device AI models (WebAssembly) in an offscreen document.
• Host: huggingface.co, hf.co, hf-mirror.com. Download the AI model files, once. Download only.
• Host: api.groq.com (and, on request, api.openai.com, api.anthropic.com, generativelanguage.googleapis.com). Reach the cloud engine you chose, with your own key; requested only when you select that provider.
• Host: noterva.com. Create or sign in to your account and validate your Premium licence on this device.
• Host: ajayceindia.com. One-time check that an offline activation code has not already been used elsewhere.

Storage, retention & security

Local data — your settings, licence state, downloaded models and any API key you enter — stays in your browser until you clear it or uninstall. Server-side we keep only your account record — your email, a hashed password, your Premium licence and your linked device IDs — for as long as your account is active. We never store your selected text or your API keys. Transfers use HTTPS. The extension contains no private keys or secrets.

Children

This is a tool for researchers and is not directed at children under 13 (or the age of digital consent in your jurisdiction).

Your rights & DPDP

We handle personal data in line with India’s Digital Personal Data Protection (DPDP) Act and comparable principles elsewhere. You may ask us to show you, correct, or delete any data associated with your account or licence by writing to us. Payment is handled separately on our website — the extension never sees card or bank details.

Changes

If this policy changes materially we will update the effective date above and, where appropriate, note it in the extension or store listing.

Contact

Questions, access requests and deletion requests: connect@noterva.com.