DuckDuckGo has announced new voice chat capabilities for Duck.ai, the private AI chatbot platform it launched last March.
The new feature lets users have real-time voice conversations with large language models (LLMs) through an encrypted relay connection. OpenAI provides the LLMs that power voice chats, but DuckDuckGo says that neither it nor OpenAI stores any audio after conversations end.
"Voice chats in Duck.ai are private, anonymized by us, and never used to train the underlying AI," the company said in its announcement. OpenAI is also contractually limited in how it can use voice chat data. Here's how DuckDuckGo describes the service:
DuckDuckGo protects your audio stream and voice data in several ways, to ensure it can only be accessed by OpenAI, (the model provider for voice chats) and only for the purpose of responding to your prompts:
- Ephemeral processing: Your live audio is streamed only while you're speaking. Once the session ends, neither DuckDuckGo nor OpenAI keeps recordings of your voice.
- No training: Your audio and the model's responses are not used to train AI models.
- Secure connection: Audio streams are encrypted in transit through WebRTC and our relay server.
- Zero data retention: Neither DuckDuckGo nor OpenAI retains any conversation data after you end a voice chat.
The feature works in most browsers except for Firefox, for which support is still in the works, according to the company. Voice chat is free within daily limits and doesn't require an account, while subscribers ($10/month) get higher daily limits in addition to DuckDuckGo VPN access, Personal Information Removal, and Identity Theft Restoration.
Users can disable voice chat at any time through Duck.ai's settings if they change their mind about using the feature. Duck.ai also offers access to non-voice models from OpenAI, Meta, Anthropic, and Mistral.
