Skip to content
KAEDEDIGITAL
עב
WhatsApp bots

WhatsApp AI chatbot policy: what your business bot may do, and how to keep it compliant

22 November 2026 9 min read Kaede Digital

In short

  • WhatsApp's business terms now distinguish a bot that serves your own customers from a general purpose AI assistant offered as the product.
  • A bot that answers questions about your business, books appointments and qualifies leads is the use most guides describe as allowed.
  • Define the bot's scope in writing, and teach it to politely decline anything outside your business.
  • A clear, fast handoff to a person keeps customers happy and keeps the bot inside its job.
  • Log every conversation and handoff in your CRM so you can review what the bot actually does.

The WhatsApp AI chatbot policy separates two kinds of bots. The first serves your own customers. The second is a general purpose AI assistant that is offered to people as the product itself. Meta’s terms for the WhatsApp Business Platform now deal with AI chatbots directly, and many owners have read the headlines as “AI bots are banned on WhatsApp”. For most small businesses, that is not what the rule is about.

The confusion is real. Several recent guides carry titles like “Not All Chatbots Are Banned”, which shows how many owners are worried. This article explains the difference in plain language. It then shows how to design scope, handoff to a person and CRM logging so your bot stays clearly on the right side of the line.

One note before we start. Platform terms are worded carefully and can change. Treat this guide as a practical explanation of the principle, not as legal advice. Before you launch, review the current terms with your WhatsApp provider.

What is the WhatsApp AI chatbot policy about?

It is about what the AI is for, not whether AI is involved at all. The terms target general purpose AI assistants: chatbots whose main job is to be an all-purpose AI that answers anything for anyone, delivered through WhatsApp.

A business bot is different. It uses AI as a tool inside a service conversation between your business and your own customers. The customer messages a dental clinic to ask about whitening, and the bot answers from the clinic’s own information. That conversation exists because of the business, not because of the AI.

Two bots, side by side

Question Business bot General purpose assistant
Why does the user message it? To get something from your business To use an AI for anything
What does it know? Your services, policies, hours, orders Whatever the user asks about
What is the product? Your service or product The AI itself

If your bot sits in the middle column, you are building the kind of bot most guides describe as allowed. Your job is to make sure it stays there.

What may a business bot on WhatsApp do?

A business bot may do almost everything a good receptionist, salesperson or support agent does in a chat, as long as it relates to your business. That covers a lot of useful work.

Typical tasks include:

  • Answering questions about your services, prices you publish, opening hours and location
  • Qualifying leads with a few relevant questions
  • Booking, moving and cancelling appointments
  • Showing order status and sending payment links
  • Browsing a catalog and recommending products you sell
  • Sending reminders, and broadcasts to customers who opted in
  • Handing the conversation to a person when needed

AI makes these tasks feel natural. Customers can write in their own words, and the bot understands. It can be trained on your own price list, FAQ and policies, so the answers are specific instead of generic.

For example, a beauty salon’s bot can explain the difference between two treatments, check which stylist is free on Thursday and book the slot. None of that turns it into a general assistant. Everything it does serves the salon’s customers. Our complete guide to WhatsApp bots for business covers these use cases in more depth.

Why do so many businesses think their bot is banned?

Mostly because the same words describe very different things. A business bot and a general purpose assistant are both called “AI chatbots”. Headlines about one sound like they apply to the other.

There is also a technical overlap. Both kinds of bot may run on similar AI models behind the scenes. Owners hear that a big AI assistant can no longer operate a certain way on WhatsApp, and they assume their own AI-powered support bot is affected too.

The model you use is not the main question. What matters is what the bot does, who it serves and what it is allowed to talk about. That is something you control through design.

How to keep your bot inside its scope

You keep a bot compliant by deciding, in writing, what it is for and then building it so it cannot drift far from that. A clear scope also makes a better bot, because focused answers are more useful than clever ones.

  1. Write the bot’s job in one sentence. For example: “This bot helps customers of our clinic get information, book appointments and reach the team.”
  2. List the topics it may cover. Services, prices, hours, bookings, orders, delivery, returns. Anything not on the list is out of scope.
  3. Train it only on your own information. Use your website, price list, policies and FAQ as its knowledge. Do not invite it to answer from general knowledge on any subject.
  4. Write a polite decline. When a customer asks for something unrelated, the bot should say so kindly and steer back: “I can help with bookings and questions about our treatments. Would you like to see available times?”
  5. Limit what it can do. Give it access only to the actions it needs, such as booking or checking an order, and nothing broader.
  6. Test it with odd requests. Ask it to write a poem, solve homework or give legal advice. Check that it declines and redirects every time.

Here is a real-life style example. A customer of an online store asks the shop’s bot to help write a cover letter. A loosely configured bot might happily do it. A well-scoped bot answers: “I’m the shop assistant, so I can’t help with that, but I can help you find a product or check your order.” The customer is not annoyed, and the bot stays in its role.

When should the bot hand off to a person?

Whenever the conversation needs judgement, empathy or a decision the bot is not allowed to make. A clear handoff is good service, and it also keeps the bot within its job.

Common handoff moments:

  • The customer asks to speak to a person
  • A complaint, refund or anything emotional comes up
  • The question is outside the bot’s scope but still about your business
  • A high-value lead is ready to talk to sales
  • The bot is not confident about the answer

The handoff should feel smooth. The person who takes over sees the whole conversation and does not ask the same questions again. The customer is told what happens next and roughly when. We go deeper on this balance in AI customer service with a human touch.

Why CRM logging matters for a compliant bot

Logging turns your bot from a black box into something you can check. If every conversation, decline and handoff lands in your CRM, you can see what the bot actually does, not just what you hoped it would do.

Useful things to record:

  • The full conversation, linked to the customer’s record
  • A tag when the bot declined an out-of-scope request
  • Every handoff, who received it and how fast they replied
  • Opt-in status for broadcasts and reminders

Review these logs regularly. If you see many out-of-scope requests on one theme, decide on purpose whether that topic belongs to your business. If it does, add it to the bot’s scope. If it does not, improve the decline message.

For example, a B2B service company reviews its bot logs and notices that prospects often ask about integration with a specific accounting tool. That is a real business question, so the team adds a short answer and a handoff to sales. The bot becomes more useful without becoming more general.

How the WhatsApp AI chatbot policy affects the setup you choose

The terms discussed here apply to the WhatsApp Business Platform, the official API that business bots run on. If you are still deciding between the free app and the API, read our comparison of the WhatsApp Business API and the Business app first.

On the platform, your provider is part of your compliance picture. Ask them three questions. How do they read the current AI terms? What controls do they offer for scope and handoff? How do conversation logs reach your CRM?

At Kaede Digital we design every bot around a written scope. It is trained on the business’s own information, has a clear path to a person and connects to the CRM from day one.

A compliance checklist for your WhatsApp bot

  • The bot’s job is written in one sentence
  • Allowed topics are listed, and everything else is declined politely
  • Knowledge comes from your own business information
  • The bot can only take the actions it needs
  • Customers can reach a person easily
  • Every conversation and handoff is logged in the CRM
  • Broadcasts go only to customers who opted in
  • Logs are reviewed regularly, and scope is updated on purpose
  • Current platform terms are checked with your provider before launch

A focused bot is a compliant bot

The WhatsApp AI chatbot policy is not a reason to avoid AI on WhatsApp. It is a reason to be clear about what your bot is for. A bot that serves your own customers, stays within a defined scope, hands off to people when needed and logs everything to your CRM is both easier to defend and better for your customers. If you want to build one that way, our team creates WhatsApp bots for business with scope, handoff and CRM connection built in from the start.

FAQ

Is it still allowed to use AI in a WhatsApp business bot?

The policy is aimed at general purpose AI assistants, not at businesses that use AI to serve their own customers. A bot that answers questions about your business, books appointments and qualifies leads is the use most guides describe as allowed. Check the current terms with your provider before launch.

What counts as a general purpose AI assistant?

It is a chatbot whose main purpose is to be an all-purpose AI that answers anything for anyone, with the AI itself as the product. A business bot is different, because people message it to get something from that specific business.

What should my bot say when a customer asks something unrelated?

It should decline politely and steer back to what it can do, for example bookings or product questions. A short, friendly redirect keeps the bot in scope without making the customer feel ignored.

Do I need to log bot conversations in a CRM?

It is not only about rules, it is good practice. Logging every conversation, decline and handoff lets you see what the bot really does, fix gaps and show that it stays within its job. It also gives your team full context when a person takes over.

WhatsApp bots

Want us to build this for you?

See the service