WhatsApp exposed the instructions it gives its AI support chatbot in the code of about 80 help and FAQ pages, according to WIRED, citing findings from privacy company Redact. On the page about deleting an account, the chatbot was told not to suggest contacting a data protection authority or any other regulator. Other instructions told it not to cite WhatsApp’s terms of service or privacy policy and not to discuss laws like GDPR, even when a user brings them up, and to avoid any statement about WhatsApp’s liability if a user mentions suing. WhatsApp removed the text after WIRED contacted Meta, and spokesperson Joshua Breckman blamed a bug, saying the instructions keep the chatbot focused on WhatsApp support rather than legal advice.
