WhatsApp Removes 100 Hidden Chatbot Prompts After Bug Exposed Them on 80 Help Pages
Updated
Updated · WIRED · Oct 19
WhatsApp Removes 100 Hidden Chatbot Prompts After Bug Exposed Them on 80 Help Pages
3 articles · Updated · WIRED · Oct 19
Summary
Around 100 hidden text blocks meant for Meta’s support chatbot were exposed in the code of roughly 80 WhatsApp help and FAQ pages, then removed after WIRED asked Meta about them.
Meta said a bug caused the display and has now been fixed; the prompts were designed to guide WhatsApp’s AI support bot toward accurate answers rather than legal advice.
The leaked instructions included “WHAT NOT TO SAY” sections telling the bot not to cite WhatsApp legal documents, discuss liability, interpret privacy laws such as GDPR or CCPA, or suggest contacting regulators.
Redact, the privacy firm that found the issue, said the exposure highlighted a gap between what users see and what LLMs are told; it remains unclear how long the text was visible or whether third-party chatbots could access it.