Anthropic Requests Claude Users to Share Voice Data for AI Training
Voluntary program raises privacy considerations for conversational and biometric data
Key Takeaways
- Anthropic is requesting voluntary voice data from Claude users for AI model training.
- Participation is opt-in; users can manage preferences via the Claude interface.
- Voice data may include biometric and sensitive conversational information.
- No known security vulnerability; the request pertains to data collection for training.
- Anthropic has not detailed retention, storage, or de-identification practices for recorded voices.
Quick answers
- What happened?
- Anthropic has initiated a program inviting Claude users to voluntarily share voice conversations to aid in the training and improvement of its AI models. The request frames participation as optional, but security analysts note the collection of voice data introduces potential privacy implications, including the creation of biometric datasets from user interactions.
- Which products are affected?
- Claude
- What should defenders do?
- Users can opt out of the voice data sharing program through the Claude privacy settings. Anthropic has not published a patch, as this is a data collection initiative rather than a software flaw.
Anthropic has started asking Claude users to voluntarily share their voice conversations to help train and improve its AI models, the company announced. The request is presented as an opt-in program, meaning users are not compelled to participate, but those who consent allow their voice recordings to be used for model training purposes. The initiative underscores the growing trend of AI developers seeking real-world conversational data to refine speech recognition, natural language understanding, and voice synthesis capabilities. While Anthropic states that participation is voluntary and users can manage their preferences through the Claude interface, privacy experts have flagged that voice data can contain biometric identifiers and sensitive personal information. The company has not disclosed specific retention or de-identification procedures, leaving unanswered questions about how the collected recordings will be stored, processed, and protected. Users concerned about data collection can review their privacy settings within the Claude platform to opt out of the program.
Security Details
This request is voluntary and does not exploit a software vulnerability. The primary concern is the collection of voice recordings, which may contain biometric identifiers and sensitive personal information. Users are advised to review privacy settings in the Claude interface to opt out if desired.
Affected products
Claude
Mitigation
Users can opt out of the voice data sharing program through the Claude privacy settings. Anthropic has not published a patch, as this is a data collection initiative rather than a software flaw.
Sources
BleepingComputer
Anthropic asks Claude users to share voice data for AI model training
Oct 4, 2026 · 10:53
Original link
Related Security News
Google Announces Gemini 4 Argon AI Model for Trusted Cyber Defenders
Google has announced Gemini 4 Argon, a new frontier artificial intelligence model being distributed to a set of trusted cyber defenders through the Fairwind Program. The company states the model delivers frontier performance across complex workflows including software engineering, enterprise knowledge work, and cybersecurity defense. Google also indicated plans to release a guardrail-free version of the model in the future. The announcement was made on October 1, 2026, via The Hacker News. No known exploitation or security vulnerabilities have been reported, as the model is currently limited to a trusted defender program and is not publicly released.
