WhatsApp's help pages hid instructions telling its support chatbot not to point users toward privacy regulators. The text was invisible to users until it was quietly removed on October 18.
Picture this. You're trying to delete your WhatsApp account, the app won't let you, and you go looking for help. You land on WhatsApp's own help pages, ask the support chatbot what to do, and it calmly steers you away from the one option your privacy policy actually guarantees: complaining to a data protection regulator.
That's not a glitch. According to reporting on the matter, WhatsApp's help pages carried hidden instructions buried in the page code, telling its support chatbot not to recommend a data protection authority to users who could no longer delete their accounts. The text was never shown on screen. It just sat there, invisible, quietly shaping the answers people got.
It was removed by October 18. But the question of how long it was there, and who put it there, is the part worth sitting with.
### What the Hidden Text Actually Did
The instructions lived in the code of the help pages, which means a regular user would never stumble across them. You'd have to open up the page source to even know they existed. Meanwhile, the chatbot reading those instructions would simply... not mention the regulator option.
Here's the awkward part. WhatsApp's own privacy policy for Europe tells users they have the right to lodge a complaint with a supervisory authority. So the public-facing promise and the hidden chatbot behavior were telling two different stories.
### Why This Matters Beyond One Chatbot
You might be thinking, so what? It's a help bot. But think about how much of your daily life now runs through automated assistants that answer questions on behalf of companies. When those assistants are quietly steered away from certain answers, you're not getting neutral help anymore. You're getting help with a leash on it.
And for anyone who works in privacy, fraud prevention, or account management, this is a reminder that AI tools are only as honest as the instructions behind them.
> "A chatbot that's told what not to say is still saying something. It's just saying it by omission."
That's the real issue. The omission is the message.
### The Practical Takeaway
If you hit a wall with any platform and the support bot seems to be going in circles, don't assume the bot has told you everything. A few habits worth keeping:
- Read the company's privacy policy yourself, especially the complaint and regulator sections.
- Ask the bot directly about your rights, then verify the answer somewhere official.
- Keep a record of what the bot told you and when. Screenshots are cheap insurance.
- When a platform removes hidden instructions after being caught, that's a signal to pay closer attention, not less.
Whether the hidden text was a deliberate choice or an oversight is still unclear. WhatsApp hasn't offered much beyond confirming the removal. But the fact that it existed at all says plenty about how carefully these automated help systems are being built and audited.
The next time a support chatbot gives you a tidy, confident answer, it's worth asking a simple question: what isn't it telling me?
