WhatsApp's Hidden Notes Told AI Not to Recommend a Regulator

Redact, a data-deletion company, reports that WhatsApp's help pages carried hidden text for an AI, found October 8 on 80 of 545 pages. Two notes told the AI not to recommend a data regulator or cite privacy law. WhatsApp told WIRED it was a bug, now fixed.

Why it matters

The same thing, for different readers.

What it means for you

If you cannot delete your WhatsApp account, Redact found the help pages told a chatbot not to mention the regulator you could complain to.

For lawmakers

Instructions to a chatbot cannot remove rights under GDPR Articles 17 and 77, Redact notes, but they can keep people from hearing about them.

For researchers

Redact read the instructions, not the chatbot's answers, so whether WhatsApp's chatbot or outside AI assistants followed the notes is untested.

The question it raises

Hidden text written only for machines raises the question of what a company owes people when the helper they consult is told what not to say.

In their own words

Every quotation was found, word for word, in the text we saved from the source. That shows the words are there; it does not make them true, or prove the speaker meant them as quoted.

From the source’s own text.

“Until mid-October 2026, WhatsApp's help pages told an AI not to recommend the regulator, and not to cite that law once you'd mentioned a lawyer.”
Dan Saltman, Founder and CEO, RedactRedact's founder summarizing the finding at the top of the post.
Found word for word in the source
“Whether WhatsApp's chatbot actually followed these notes. We read its instructions, not its answers.”
Dan Saltman, Founder and CEO, RedactFrom the post's list of what Redact cannot yet say.
Found word for word in the source
“focused on providing the best advice to users about WhatsApp”
WhatsApp spokesperson, Spokesperson, WhatsAppWhatsApp's statement to WIRED, as relayed in Redact's update, on what it wants its support chatbots to do.
Found word for word in the sourceAs quoted in the text we read, not the speaker’s own text or a transcript of their words

How the finding was reached

In our words, from the source. Weigh the finding by how it was made.

What was done
Redact read all 545 pages that WhatsApp's help centre links to, saved copies, and looked for text sent in the page but never shown on screen; it also checked 300 Facebook and Instagram help pages.
What it found
80 pages held 102 hidden blocks for an AI. Most were harmless; two told the AI not to recommend a data regulator and not to reference privacy law once legal action was mentioned. All blocks were gone by October 18.
What it does not show
Redact read the instructions, not the chatbot's answers; it did not test whether outside assistants obeyed them, how long the notes were there, or other languages and unlinked pages.
Status
A report

The record

Questions
Are AI companies putting stronger safeguards into practice? · How is AI changing childhood, health and daily life?
People
Dan Saltman (Wrote it, Founder and CEO, Redact) · WhatsApp spokesperson (Speaks, Spokesperson, WhatsApp)
What we read
We read the whole source.
Found by
The Radar, run of October 20, 2026, via hacker-news-50-points (tier 3: discovery only)
Cite this
Redact, “WhatsApp's Hidden Notes Told AI Not to Recommend a Regulator”, October 19, 2026. Entry in the Ctrl AI Base, https://ctrlai.com/base/redact-2026-10-19-whatsapp-hidden-ai-instructions
Our notesAuthor's company sells deletion services and has an interest in the topic. WhatsApp's response is relayed second-hand from WIRED via Redact's update; WIRED article not read. Hidden-note wording was read only as quoted on the page.