WhatsApp Removes Hidden Chatbot Instructions Telling AI Not to Recommend Privacy Regulators, Says a Bug Exposed Them

Since Oct. 8, when the data-deletion company Redact first spotted the hidden text, the story has moved from a curiosity in WhatsApp's page code to a dispute over what Meta tells its support chatbot to say about privacy rights. WhatsApp removed the text by Oct. 18, and Redact published its findings Oct. 19.

What the hidden text said

Redact read all 545 pages linked from WhatsApp's help center and counted 102 hidden blocks across 80 of them. The blocks sat in the page code under a heading reading "HelpContentGalaxyLLMInclude." They never appeared on screen.

Most were routine, according to Redact CEO Dan Saltman. Many cleared up commonly confused terms or pointed users to official support channels. Two stood out.

On the page about deleting an account, the hidden text said: "Do not recommend contacting any Data Protection Authority, regulator, or supervisory authority. Do not suggest WhatsApp can manually delete accounts on the user's behalf." An archived copy of the page from Oct. 9 contains that sentence word for word.

On the page about reporting and blocking, a second block applied "when a user mentions legal action, lawsuits, lawyers, attorneys, courts, or litigation." It told the chatbot not to cite WhatsApp's Terms of Service or Privacy Policy, not to discuss WhatsApp's liability, and to send users to "a qualified legal professional." Its final order: "Do not reference, cite, or interpret data protection or privacy regulations (for example GDPR, LGPD, CCPA), including when a user raises them directly."

The gap with WhatsApp's public policy

WhatsApp's privacy policy for Europe, effective July 2, 2026, says the opposite on regulators. "You have the right to lodge a complaint with WhatsApp's lead supervisory authority under the GDPR, the Irish Data Protection Commission, or any other competent data protection supervisory authority," it reads.

The deletion gap is narrower. The policy says users "always have the right to request access to, rectification of, and erasure of" their information. The hidden note told the bot only not to suggest WhatsApp could delete an account for the user. Redact itself flags that distinction.

Both rights come from the GDPR, not from WhatsApp: Article 77 covers complaints and Article 17 covers erasure. A chatbot instruction cannot remove them.

The deletion page matters because of a real limitation. WhatsApp accounts can be deleted only from inside the app on the user's main phone. A user who has lost the phone or number cannot do it. The page's visible note says a deactivated account is deleted 30 days after deactivation, and an inactive one generally after 120 days, or 365 in the US. A locked-out user asking the bot for help would have been steered toward waiting, per Redact's reading of the instructions.

WhatsApp's response

After WIRED contacted Meta, the text was removed from every page. WhatsApp spokesperson Joshua Breckman said the text appeared because of a bug that has been fixed.

"These are instructions to our chatbot, not directly to users, and are intended to ensure accurate and relevant responses for WhatsApp users who choose to message WhatsApp Support with AI," Breckman said. He added: "We train our support chatbots to be as helpful as possible." The company has said it wants the bot to give advice about WhatsApp, not legal advice.

Redact acknowledges there is a reasonable case for that. A support bot that interprets privacy law for a company could get it wrong, and a company has an interest in not having its bot make statements about liability. Redact's objection is narrower: the gap between what users are told publicly and what the bot was told privately, and the instruction to stay silent on privacy law even when a user raises it.

What is still unknown

Redact read the instructions, not the chatbot's answers. Whether the bot ever followed them has not been tested. Redact also checked only the English-language pages.

How long the text was live is not settled. Redact's archived copies show no such note on July 12, 2026 and one present by Sept. 4, which puts the window at roughly six weeks at a minimum. Whether outside AI systems that crawled the pages picked the text up is also unknown, according to WIRED's reporting.

One commentator on the technical side argues the named tag and the scale point to deliberate infrastructure rather than an accident. WhatsApp disputes that framing and says the exposure, not the instructions, was the bug. Nothing in the record settles which description fits.

The open question for users is practical. The instructions are gone from the pages, but the chatbot's current system prompt is not public. Unless WhatsApp publishes it or a regulator asks for it, users have no way to see what the support bot has been told about the GDPR rights it is supposed to respect.