Anthropic asks Claude users to share voice data for AI model training

A major tech firm is asking its users to share their voice conversations with them in order to improve their AI models, raising important questions about user consent and data protection.

Anthropic, a company behind the popular AI assistant Claude, has started prompting users to allow it to use their voice data for training purposes. This is not an automatic process, however – users can choose whether or not to share their voice recordings with Anthropic. The prompt appears when using Claude’s voice features and explicitly asks for permission.

It’s worth noting that this is a separate option from Anthropic’s existing feature that allows the company to use chat data and coding sessions for model training. This means that users have more control over what data they share with Anthropic, and can choose to allow voice data sharing while keeping other types of data private.

When inside Claude’s Privacy settings, users will find a dedicated toggle labeled “Allow us to use your voice data”. By default, this option is disabled, so Anthropic isn’t automatically opting users into sharing their voice recordings. Users can also turn the option off later or delete the voice data altogether if they change their mind.

This development highlights an important issue in AI research: the need for high-quality training data to improve model performance. While it’s understandable that companies like Anthropic would want access to user-generated data, users have a right to know how their information is being used and to control what data is shared with them.

As the use of AI continues to grow, so does the risk of data misuse and exploitation. It’s essential for both developers and users to be aware of these issues and to take steps to protect sensitive information.

In practical terms, this means paying close attention to prompts like Anthropic’s voice data sharing request, and thinking carefully about whether or not to allow access to sensitive data. By being informed and taking control of our own data, we can help ensure that AI research benefits everyone – not just the companies driving it forward.


Source: Bleeping Computer — 2026-10-04