The Pulse
Anthropic Adds a Separate Voice-Data Training Choice in Claude
Claude users can choose separately whether Anthropic may use audio recordings and voice-chat data to improve its models. The voice-data setting is distinct from the existing choice for chats and coding sessions.
AI.info Team · Written with AI ·
Claude asks before using voice sessions
Anthropic is asking Claude users whether it may use their voice data to improve its AI models, BleepingComputer reported after seeing the interface. The request appears when users access Claude’s voice features. It offers “Allow” and “Not now,” rather than silently enrolling users, and the report says the setting was off by default in the interface it examined.
The prompt says audio recordings and voice-chat data can help Anthropic’s models understand and respond to speech. It also tells users they can change the choice or delete the data in settings. The wording makes the request specifically about audio, not simply the text of a conversation conducted by voice.
A separate switch from chat training
Anthropic places the voice-data choice in Claude’s Privacy settings, under the description “Allow us to use your voice data.” The setting says it applies to “approved sessions,” but the report does not explain what qualifies a session for approval or how long voice data is kept. Users can reportedly turn the option off later and delete voice data through settings.
The new control is separate from the existing choice about using chats and coding sessions to improve Claude. BleepingComputer reporter Mayank Parmar writes:
“That means you could allow Anthropic to train on your voice conversations while keeping regular chats and Claude Code sessions excluded, or the other way around.”
That separation lets users make different choices for voice conversations and text or coding activity, rather than treating all Claude interactions as one data category. The distinction matters for people who use voice mode for personal or sensitive discussions but want different settings for other work.
Voice mode has expanded across Claude
In a July 23, 2026 product announcement, Anthropic said voice mode could use Claude Opus, Sonnet and Haiku, connect to user-authorized tools, and support additional languages. Anthropic said the beta was available to chat users on mobile, desktop and web.
That broader availability means more kinds of conversations may happen aloud, including longer sessions in which users think through work or personal decisions. Anthropic’s prompt describes recordings as well as voice-chat data, but the report does not specify whether the recordings include only a user’s speech, what happens to audio after processing, or whether the new choice applies identically across plans and platforms.
Anthropic’s published policy leaves voice details open
Anthropic’s consumer privacy guidance says the company may use chats and coding sessions for model improvement when users allow it, when conversations are flagged for safety review, or when users otherwise opt in. The guidance does not spell out the new voice-specific option or its retention terms.
Claude users can check Settings > Privacy and make the voice choice independently of their chat and coding preferences. Anyone who previously enabled model improvement for text should not assume that setting also decides the separate voice-data choice. The interface, as described by BleepingComputer, presents voice sharing as optional and provides a way to reverse the choice.