Hub 4 AI Voice Conversation Feature Privacy Policy

Please read this Policy in addition to the SwitchBot App Privacy Policy, which you have agreed to. This Policy explains how we collect, transfer, use, and store your personal information when you use the Hub 4 AI Voice Conversation Feature (the “Feature”).

When you enable “AI Voice Conversation” in the App, we will present this Policy to you. The Feature will be enabled only after you tap “Agree.” If you tap “Cancel” or do not agree, the Feature will remain disabled, and we will not collect or process your voice information through the Feature. If you disable the Feature, you will not be able to use voice interaction or conversation history, but the normal operation of Hub 4 and other basic App features will not be affected. “Conversation History” is optional. If you do not enable it, you can still use real-time voice interaction, but the system will not save your conversation history.

1. Information We Collect or Process

When you enable and use this Feature, we may collect or process the following information to recognize voice commands, generate responses, and control devices:

  • Voice information: Real-time voice input captured by the Hub 4 microphone;
  • Conversation information: Text converted from voice, system responses, and contextual information needed to understand the conversation;
  • Device interaction information: Names of devices you ask to control or query, command content, device status, and operation results;
  • Location information: The city configured in Home Management or a location you specify in the conversation when you use features such as weather queries;
  • Service operation information: De-identified operational data, such as conversation duration and response time, used to maintain service stability, troubleshoot issues, and improve the user experience.

2. Transfer, Use, and Storage of Information

After you issue a voice command, the relevant information is processed as follows: Hub 4 transmits real-time voice through an encrypted channel to a third-party cloud-based speech recognition provider to convert the voice into text. After speech recognition is complete, the converted text, together with the city or location information necessary to provide weather queries, is transmitted to a third-party AI large language model provider to analyze the command and generate a response. The third-party AI large language model provider does not receive the original voice. The generated response is returned to Hub 4 and the App to execute commands, control devices, or display interaction results.

The third-party providers mentioned above process relevant information solely to provide this Feature, and processing takes place at server nodes corresponding to your region. They will not transfer the relevant information to other regions. We will not use your original voice, recognized text, or conversation content to train AI models, nor will we provide it to third parties other than the providers mentioned above, unless otherwise required by law or you separately consent.

After each real-time voice interaction is completed, the original voice is deleted as soon as speech-to-text processing is complete. The session ends without any history being saved. Only when you enable or use “Conversation History” will the processed text and responses be saved on our cloud servers. In the App, they will be grouped by session and listed by date for you to view. They will be retained for no longer than 30 days. Even when “Conversation History” is enabled, the cloud servers will not retain the original voice or provide voice playback.

We retain relevant information only for as long as necessary to achieve the purposes of this Feature. When the retention period expires, the processing purpose has been fulfilled, you delete the relevant data, or we discontinue the relevant Feature, we will delete or anonymize the personal information in accordance with applicable law, unless otherwise required by law.