WhatsApp Launches 'Writing Help' AI as Experts Urge Caution After Security Audits
The optional English-first tool uses a privacy-focused processing system, with Meta saying issues found in reviews were fixed before rollout.
Overview
- Writing Help offers AI-generated rewrites with selectable tones such as factual, professional, humorous or empathetic, and users can accept or edit suggestions.
- The feature is opt-in and must be enabled in settings; once activated, a new pen icon appears in chats to access the assistant.
- Initial availability is in English for users in the United States and select other countries, with broader language and market expansion planned later this year.
- WhatsApp says requests are anonymized through redirectors, processed in isolated server enclaves and deleted immediately, preventing Meta from viewing message content.
- Independent audits by NCC Group and Trail of Bits reported 49 potential vulnerabilities in early tests; Meta states the findings were addressed before launch, while the researchers still advise caution and call for greater transparency, noting the feature remains off by default.