At 02:08 UTC on October 18, 2026, Redact found all 102 hidden instruction blocks absent from the 80 WhatsApp Help Center pages it had reviewed. The pages themselves remained online.
What Redact found in WhatsApp’s Help Center
Redact says researcher Dan Saltman reviewed 545 English-language pages linked from WhatsApp’s Help Center. It reported hidden text on 80 pages, in 102 blocks intended for an AI support chatbot.
One block on an account-deletion page directed the chatbot not to recommend contacting a data-protection authority or suggest that WhatsApp could manually delete an account for the user. Another directed it not to cite privacy regulations, including GDPR, LGPD and CCPA, when users raised legal action, lawyers or courts.
WhatsApp’s explanation and the dated checks
WhatsApp told WIRED that the text was intended for its support chatbot and appeared because of a bug. Redact says it found the hidden text on October 8 and began reviewing the linked pages.
For the account-deletion page, Redact says an archived copy from July 12, 2026, did not contain the note, while a September 4 copy did. Redact’s October 15 check, at 21:40 UTC, found both example notes still present. At 02:08 UTC on October 18, its check found all 102 blocks absent from the 80 pages it had reviewed.
What the findings say about chatbot privacy
Redact did not test whether WhatsApp’s chatbot or any outside AI system followed the instructions. Its review covered English-language pages linked from the Help Center. WhatsApp says personal messages are end-to-end encrypted and that Meta AI can access only messages people choose to send it.
