Talk to an expert
PricingNewsTalk to an expert
Back to Newsroom
Product

Anthropic Adds Opt-In Voice Data Training Toggle to Claude

Anthropic is now asking Claude users for explicit permission before using their voice conversations to train future AI models, through a new consent prompt and a dedicated privacy toggle.

4T4Tify Team4Tify
Oct 4, 2026·3 min read
Anthropic Adds Opt-In Voice Data Training Toggle to Claude

Anthropic Adds Opt-In Voice Data Training Toggle to Claude

Anthropic has rolled out a new consent prompt that asks Claude users whether they want their voice conversations used to train and improve the assistant's AI models.

What happened

When using Claude's voice features, users now see a dialog explaining that audio recordings and voice chat data can help the models understand and respond to speech better. The choice is presented as a simple Allow or Not now, and a matching toggle lives in Settings > Privacy under "Allow us to use your voice data." In the version currently rolling out, the setting is off by default — Anthropic is not automatically enrolling anyone into voice-data training.

Importantly, this permission is kept separate from Anthropic's existing "Help improve our AI models" toggle, which governs text chats and Claude Code sessions. That separation means a user could opt in to voice training while keeping text conversations excluded from training, or the reverse — the two data streams are controlled independently.

Why it matters

Voice data is a more sensitive category than typed text: it can carry biometric characteristics, background context, and incidental information about other people in the room. Giving users a standalone, granular control — rather than bundling voice consent into a general "help us improve" switch — is a meaningful privacy design choice, and it reflects growing regulatory and user expectations around how AI vendors handle spoken data specifically.

For organizations that have approved Claude for employee or customer-facing use, this is also a reminder that AI assistant privacy settings are not static — vendors periodically introduce new data-collection categories that need to be reviewed against internal data-handling policies.

What to do

  • Check Settings > Privacy in Claude and confirm the "Allow us to use your voice data" toggle matches your organization's policy before anyone uses voice features.
  • Treat this as a distinct control from chat/Claude Code training settings — review both independently.
  • If voice data was enabled unintentionally, Anthropic states it can be deleted from the same settings panel.
  • In regulated environments, document the current default (currently off) and revisit it periodically, since defaults and prompts can change with future product updates.
SHARE