Privacy Policy — Voice AI Extension

Last updated: 17 May 2026 · Applies to: Voice AI browser extension v0.1.x (Chrome / Firefox) · Source: github.com/Smolevich/voice-ai-extension

Summary

The Voice AI extension reads selected text aloud. It runs entirely on your device. The authors do not operate any server that this extension talks to. There is no account, no telemetry, no analytics, no advertising.

What the extension stores

Your preferences (voice source, selected voice, speed, pitch) — kept in chrome.storage.sync so they follow your browser profile.
An optional Groq API key, if you choose to use the Groq voice source — also kept in chrome.storage.sync. The key never leaves your device except when it is sent to api.groq.com to request synthesis, exactly as it would be if you used your Groq key directly.

chrome.storage.sync is encrypted by the browser vendor (Google for Chrome, Mozilla for Firefox) and synced across your signed-in browser sessions if you have sync enabled. Neither the extension authors nor any server controlled by us has access to this storage.

What the extension does NOT collect

No account, sign-in, or registration of any kind.
No telemetry, analytics, crash reports, or usage statistics.
No advertising or marketing data.
No reading of your browsing history.
No persistent storage of selected text. The text you ask to be read aloud lives in memory only while it is being read, then is discarded.
No tracking across sessions, devices, or browser profiles.

Where data goes when you use the extension

What you do	What leaves your device
Use the Browser voice source (default)	Nothing. Synthesis happens locally in your operating system.
Use the Groq voice source with your API key	The selected text and your Groq API key are sent to `https://api.groq.com`. Groq returns audio, which the extension plays in the current tab and discards. We do not see, proxy, or log this traffic. Groq's own terms apply: groq.com/privacy-policy.

Third parties

The extension contacts api.groq.com and only when you have configured the Groq voice source. No other network endpoints are contacted by the extension code.

Permissions explained

Permission	Why
`storage`	Save your settings and optional API key locally on your device.
`activeTab`	Read the text you have selected on the current page, only when you explicitly invoke the extension via popup, context menu, or keyboard shortcut.
`scripting`	Inject the speech synthesis call into the current tab so audio plays from the page context. Triggered only by a user gesture.
`contextMenus`	Add the right-click "Speak selected text" entry.
content scripts on `http` and `https` pages	Cache the last text selection so that the right-click menu and keyboard shortcut can speak your full selection (otherwise macOS Chrome collapses the selection to a single word at right-click time). The content script only reads the selection — it never modifies pages.
`api.groq.com` host permission	Send selected text to Groq when you have explicitly chosen the Groq voice source.

Changes to this policy

If what the extension handles changes, this page will be updated and the "Last updated" date above will move. Material changes (new providers, future Premium tier hitting an authors' server, any telemetry) will also appear in that release's notes on GitHub.

Future Premium tier

A managed Premium subscription is on the roadmap. When it ships, this policy will gain a section describing what data flows to the authors' server. Until then, no such tier is enabled in the released extension.

Contact

Issues and questions: github.com/Smolevich/voice-ai-extension/issues
Email: support@voice.smolevich.com