WhatsApp's help pages leaked hidden instructions for a Meta chatbot, Wired reports, offering an early look at how the company intends to wire generative AI into one of its largest user surfaces.
Instructions buried in support documentation tend to read like prompts: they define the assistant's role, tone, and boundaries before the feature is officially announced. That makes the leak relevant to users, who get a preview of how AI will sit inside the messaging layer used by billions, and to developers, who rarely get to see the instruction text behind a consumer chatbot at this scale.
The episode is also a textbook gap between development and publication: material belonging to an unreleased product surfaces on a public support page and becomes an accidental teaser. WhatsApp help pages are a recurring source of such reveals, since support content often ships ahead of the feature it describes.
Open questions include which product form the instructions describe, whether they are still current, and how Meta has responded. The details are in Wired's report.
Read more: Wired.
