Skip to content

Security and privacy, by architecture

Render the avatar on your own hardware and its audio and video stay there. This page lists what reaches bitHuman in each mode.

What leaves your building

  • bitHuman cloud

    Portrait
    Uploaded once to create the avatar (Reaches bitHuman)
    Live audio
    Reaches bitHuman, encrypted in transit (Reaches bitHuman)
    Avatar video
    Rendered by bitHuman, streamed to you (Reaches bitHuman)
    Usage reports
    Metered by bitHuman (Reaches bitHuman)
  • Your servers

    Portrait
    Uploaded once to create the avatar (Reaches bitHuman)
    Live audio
    Goes where the conversation runs (Your services)
    Avatar video
    Stays on your machine (Stays on your hardware)
    Usage reports
    No audio, video or conversation text (Reaches bitHuman)
  • On the device

    Portrait
    Uploaded once to create the avatar (Reaches bitHuman)
    Live audio
    Goes to your app's voice services (Your services)
    Avatar video
    Renders and stays on the device (Stays on your hardware)
    Usage reports
    Usage metering only (Reaches bitHuman)
  • Fully offline

    Portrait
    Uploaded once to create the avatar (Reaches bitHuman)
    Live audio
    Stays on the machine (Stays on your hardware)
    Avatar video
    Stays on the machine (Stays on your hardware)
    Usage reports
    Metered on the machine (Stays on your hardware)

With the web embed, the conversation runs on bitHuman's servers, even when the avatar renders in the tab.

Every data flow, in the docs

Controls

Compliance starts with where the data goes

GDPR and CCPA obligations follow the data, so we start with architecture.

In on-device, self-hosted and offline deployments, customer conversations stay in systems you already govern, and your existing programs apply.

Privacy policy

Questions

Does bitHuman train on our data?
Your photo is used to build your own avatar. If you need a written answer about model training for your organization, contact us.

Start a security review