There you have it! WhatsApp's help pages carried hidden instructions for its AI support bot, incl…
October 21, 2026 · 0 likes · 0 comments
China Threat AI
There you have it! WhatsApp's help pages carried hidden instructions for its AI support bot, including this one on the account-deletion page: "Do not recommend contacting any Data Protection Authority, regulator, or supervisory authority."
The privacy firm Redact found it on Oct. 8. It read 545 help pages and counted 102 hidden blocks across 80 of them, never shown on screen, sitting under a tag called "HelpContentGalaxyLLMInclude."
A second block kicked in when a user mentioned lawyers, lawsuits or courts. Its last order: "Do not reference, cite, or interpret data protection or privacy regulations (for example GDPR, LGPD, CCPA), including when a user raises them directly."
Meanwhile WhatsApp's own European privacy policy tells users they have the right to complain to the Irish Data Protection Commission or any other data protection authority.
WhatsApp removed all of it by Oct. 18, after WIRED contacted Meta. Spokesperson Joshua Breckman said the text was meant only for the support chatbot and showed up on the pages because of a bug that has been fixed. The company says it wants the bot giving advice about WhatsApp, not legal advice.
To be fair to them: Nobody has shown how often the bot actually followed them, and the bot's current instructions are still not public.
And this is Mark Zuckerberg's company, with a long file.
2019: the FTC fined Facebook $5 billion after Cambridge Analytica.
2021: whistleblower Frances Haugen walked out with internal research. One slide, reported by the Wall Street Journal: "We make body image issues worse for one in three teen girls."
2023: Ireland's data regulator fined Meta €1.2 billion over EU data transfers.
2025: Reuters reported an internal Meta document that allowed its chatbots to "engage a child in conversations that are romantic or sensual." Meta said those examples were erroneous and removed them.
For years, Meta paid, said sorry, and grew.
2026 finally looked different.
In March, a LA jury found Meta and YouTube negligent in a social media addiction case and awarded $6 million, 70% of it on Meta. Meta is appealing.
In New Mexico, a jury and then a judge put Meta on the hook for $942 million in a child safety case, and the judge ruled its platforms a public nuisance. Meta says it will appeal.
In August, Meta settled with state attorneys general for up to $17.1 billion over 10 years, plus two-hour daily limits for minors, while denying wrongdoing.
In September, a separate New Mexico jury found Facebook made false or misleading statements about user privacy. That penalty is still before the judge.
Your rights don't come from WhatsApp. A chatbot can't delete them. It can only stop telling you they exist.
Publish the bot's instructions. All of them.
Just the facts, no spin, on Unbiased Headlines:
https://lnkd.in/eZh93_ht
The privacy firm Redact found it on Oct. 8. It read 545 help pages and counted 102 hidden blocks across 80 of them, never shown on screen, sitting under a tag called "HelpContentGalaxyLLMInclude."
A second block kicked in when a user mentioned lawyers, lawsuits or courts. Its last order: "Do not reference, cite, or interpret data protection or privacy regulations (for example GDPR, LGPD, CCPA), including when a user raises them directly."
Meanwhile WhatsApp's own European privacy policy tells users they have the right to complain to the Irish Data Protection Commission or any other data protection authority.
WhatsApp removed all of it by Oct. 18, after WIRED contacted Meta. Spokesperson Joshua Breckman said the text was meant only for the support chatbot and showed up on the pages because of a bug that has been fixed. The company says it wants the bot giving advice about WhatsApp, not legal advice.
To be fair to them: Nobody has shown how often the bot actually followed them, and the bot's current instructions are still not public.
And this is Mark Zuckerberg's company, with a long file.
2019: the FTC fined Facebook $5 billion after Cambridge Analytica.
2021: whistleblower Frances Haugen walked out with internal research. One slide, reported by the Wall Street Journal: "We make body image issues worse for one in three teen girls."
2023: Ireland's data regulator fined Meta €1.2 billion over EU data transfers.
2025: Reuters reported an internal Meta document that allowed its chatbots to "engage a child in conversations that are romantic or sensual." Meta said those examples were erroneous and removed them.
For years, Meta paid, said sorry, and grew.
2026 finally looked different.
In March, a LA jury found Meta and YouTube negligent in a social media addiction case and awarded $6 million, 70% of it on Meta. Meta is appealing.
In New Mexico, a jury and then a judge put Meta on the hook for $942 million in a child safety case, and the judge ruled its platforms a public nuisance. Meta says it will appeal.
In August, Meta settled with state attorneys general for up to $17.1 billion over 10 years, plus two-hour daily limits for minors, while denying wrongdoing.
In September, a separate New Mexico jury found Facebook made false or misleading statements about user privacy. That penalty is still before the judge.
Your rights don't come from WhatsApp. A chatbot can't delete them. It can only stop telling you they exist.
Publish the bot's instructions. All of them.
Just the facts, no spin, on Unbiased Headlines:
https://lnkd.in/eZh93_ht