WhatsApp's Help Pages Had Hidden AI Instructions
Original: WhatsApp's help pages hid LLM instructions: dont recommend a regulator
Why This Matters
Hidden AI steering on consumer help pages raises direct questions about GDPR compliance and user rights.
WhatsApp embedded hidden AI instructions in 80 help pages, telling its support chatbot not to recommend data protection regulators or cite relevant laws. Redact found them on Oct 8, 2026; WhatsApp removed all 102 blocks by Oct 18.
Researchers at Redact discovered that WhatsApp's public help pages contained hidden text blocks—invisible to users but readable by AI—that instructed a support chatbot to withhold specific information. Two instructions stood out. On the account deletion page, the hidden text told the AI not to recommend a data protection regulator to users who couldn't delete their accounts. On a separate page, the AI was instructed not to mention the law underpinning a user's rights once they had mentioned a lawyer—despite WhatsApp's own European privacy policy naming that law and the right to file a regulator complaint.
Redact read all 545 pages linked from WhatsApp's help center and found hidden blocks on 80 of them, totaling 102 blocks. Most were described as harmless. The findings were confirmed by copies saved in the Internet Archive. WIRED reported on the discovery on October 19. WhatsApp told WIRED the notes were instructions for its AI support chatbot and appeared in the pages due to a bug, now fixed. A spokesperson said the company wants its chatbot "focused on providing the best advice to users about WhatsApp" and "not to offer legal advice." All hidden blocks were removed between October 15 and October 18, 2026.