In a significant move for privacy-conscious users, DuckDuckGo officially launched a real-time voice chat feature for its AI platform, Duck.ai, on February 9, 2026.
The feature brings a conversational layer to Duck.ai, which originally launched in early 2025. It allows you to speak directly to an AI assistant and receive instant, spoken responses—all while maintaining the strict anonymity that is DuckDuckGo’s hallmark.
Key Features of Duck.ai Voice Chat
| Feature | Description |
| Encrypted Relay | Audio is streamed through an encrypted relay that DuckDuckGo cannot decrypt. |
| Model Provider | All voice interactions are currently powered by OpenAI models. |
| Zero Data Retention | Neither DuckDuckGo nor OpenAI stores your audio or transcripts after the session ends. |
| No Training | Your voice data is contractually protected and never used to train underlying AI models. |
| Accessibility | Works on most modern browsers (except Firefox, which is “coming soon”). |
The Privacy “Catch”: Biometric Identity
While DuckDuckGo ensures that your data is not stored or linked to an account, they have been transparent about one inherent risk: your voice is a biometric identifier. > “Your unaltered voice is unique to you… it could theoretically be used to recognize you. DuckDuckGo does not attempt to identify, fingerprint, or link voices to users.” — DuckDuckGo Help Documentation
Because of this, the feature is strictly opt-in. You must manually enable it in your settings and grant microphone permissions each time you start a new session.
How to Use It
- Navigate to Duck.ai.
- Look for the “New Voice Chat” option in the left-hand navigation bar.
- Click “Enable Voice Chat” and confirm the browser microphone prompt.
- To end the session and wipe the local cache, simply click the Fire Button.
Usage Limits
- Free Users: Have access to a set number of daily voice interactions.
- DuckDuckGo Subscribers: Benefit from significantly higher usage caps and access to more advanced reasoning models.
