Home / Technology / WhatsApp's Secret Chatbot Rules Exposed by Bug
WhatsApp's Secret Chatbot Rules Exposed by Bug
19 Oct
Summary
- Around 80 WhatsApp web pages revealed chatbot instructions.
- Instructions included 'WHAT NOT TO SAY' to users.
- Meta claims a bug caused the unintentional text exposure.
Internal instructions for WhatsApp's AI chatbot, including 'WHAT NOT TO SAY' directives, were inadvertently exposed on approximately 80 help and FAQ pages. These hidden texts, found within the webpages' underlying code, offered guidance to steer the chatbot's responses and illustrated how such instructions are crafted.
Privacy company Redact discovered the exposed text, noting it advised against recommending contact with data protection authorities or providing legal advice. Specific instructions also detailed how to handle mentions of legal action, privacy regulations like GDPR, and user-specific information such as hearing impairments.
A WhatsApp spokesperson confirmed the exposure was due to a bug that has since been rectified. These instructions were intended for the chatbot to ensure accurate responses for users contacting WhatsApp Support via AI, and not for direct public view. The company asserts its commitment to providing transparent information on user data rights.
